diff --git a/DESCRIPTION b/DESCRIPTION
index ce8965b495..0d82874660 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -14,7 +14,7 @@ Authors@R: c(
   person("Hugh","Parsonage",  role="ctb"))
 Depends: R (>= 3.1.0)
 Imports: methods
-Suggests: bit64, curl, knitr, xts, nanotime, zoo, reshape2, testthat (>= 0.4)
+Suggests: bit64, curl, knitr, xts, nanotime, zoo, reshape2
 Description: Fast aggregation of large data (e.g. 100GB in RAM), fast ordered joins, fast add/modify/delete of columns by group using no copies at all, list columns, friendly and fast character-separated-value read/write. Offers a natural and flexible syntax, for faster development.
 License: MPL-2.0 | file LICENSE
 URL: http://r-datatable.com
diff --git a/inst/tests/tests.Rraw b/inst/tests/tests.Rraw
index a0821931c2..d871ad20e3 100644
--- a/inst/tests/tests.Rraw
+++ b/inst/tests/tests.Rraw
@@ -11840,6 +11840,149 @@ gc()
 after = sum(gc()[, 2])
 test(1912.2, after < before + 10)
 
+# BEGIN port of old testthat tests, #2740.  Issue numbers may be from R-forge.
+#
+# test-data.frame-like.R (merge, subset, transform)")
+#
+# `x` columns are valid, #1299
+d1 <- data.table(x=c(1,3,8), y1=rnorm(3), key="x")
+d2 <- data.table(x=c(3,8,10), y2=rnorm(3), key="x")
+ans1 <- merge(d1, d2, by="x")
+ans2 <- cbind(d1[2:3], y2=d2[1:2]$y2)
+setkey(ans2, x)
+test(1913.1, ans1, ans2)
+#
+# `xkey` column names are valid in merge, #1299
+d1 <- data.table(xkey=c(1,3,8), y1=rnorm(3), key="xkey")
+d2 <- data.table(xkey=c(3,8,10), y2=rnorm(3), key="xkey")
+ans2 <- cbind(d1[2:3], y2=d2[1:2]$y2)
+setkey(ans2, xkey)
+test(1913.2, merge(d1, d2, by="xkey"), ans2)
+#
+# one column merges work, #1241
+dt <- data.table(a=rep(1:2,each=3), b=1:6, key="a")
+y <- data.table(a=c(0,1), bb=c(10,11), key="a")
+test(1913.3, merge(y, dt), data.table(a=1L, bb=11, b=1:3, key="a"))
+test(1913.4, merge(y, dt, all=TRUE),
+             data.table(a=rep(c(0L,1L,2L),c(1,3,3)),
+                        bb=rep(c(10,11,NA_real_),c(1,3,3)),
+                        b=c(NA_integer_,1:6), key="a"))
+#
+# y with only a key column
+y <- data.table(a=c(0,1), key="a")
+test(1913.5, merge(y,dt), data.table(a=1L, b=1:3, key="a"))
+test(1913.6, merge(y, dt, all=TRUE), data.table(a=rep(c(0L,1L,2L),c(1,3,3)), b=c(NA_integer_,1:6), key="a"))
+#
+# merging data.tables is almost like merging data.frames
+d1 <- data.table(a=sample(letters, 10), b=sample(1:100, 10), key='a')
+d2 <- data.table(a=d1$a, b=sample(1:50, 10), c=rnorm(10), key='a')
+dtm <- merge(d1, d2, by='a', suffixes=c(".xx", ".yy"))
+dtm.df <- as.data.frame(dtm)
+dfm <- merge(as.data.frame(d1), as.data.frame(d2), by='a', suffixes=c('.xx', '.yy'))
+test(1913.7, unname(dtm.df), unname(dfm))
+test(1913.8, colnames(dtm), colnames(dfm))
+#
+## merge and auto-increment columns in y[x]
+## merging tables that have common column names that end in *.1 gets
+## tricky, because the y[x] mojo does some magic to increment the *.1
+## in the x (I think) and keep *.1 in the y
+x <- data.table(a=letters[1:10], b=1:10, b.1=1:10 * 10, key="a")
+y <- data.table(a=letters[1:10], b=letters[11:20], b.1=rnorm(10), key="a")
+M <- merge(x, y)
+m <- merge(as.data.frame(x), as.data.frame(y), by="a")
+test(1913.9, is.data.table(M) && !is.data.table(m))
+test(1913.11, all(names(M) %in% union(names(M), names(m))))
+for (name in names(m)) {
+  test(1913.12, M[[name]], m[[name]])
+}
+#
+# Original example that smoked out the bug
+M <- data.table(a=letters[1:10], b=1:10)
+m <- as.data.frame(M)
+ms <- lapply(1:3, function(x) data.table(a=letters[1:10], b=1:10 * 10^x))
+for (i in 1:3) {
+  M <- merge(M, ms[[i]], by='a', suffixes=c("", sprintf(".%d", i)))
+}
+for (i in 1:3) {
+  m <- merge(m, as.data.frame(ms[[i]]), by='a', suffixes=c("", sprintf(".%d", i)))
+}
+test(1913.13, is.data.table(M) && !is.data.table(m))
+test(1913.14, all(names(M) %in% union(names(M), names(m))))
+for (name in names(m)) {
+  test(1913.15, M[[name]], m[[name]])
+}
+#
+# simple subset maintains keys
+dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 c=sample(20), key='a')
+sub <- subset(dt, a == 'b')
+test(1913.16, key(dt), key(sub))
+#
+# subset using 'select' maintains key appropriately"
+dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 c=sample(20), key=c('a', 'b'))
+sub.1 <- subset(dt, a == 'a', select=c('c', 'b', 'a'))
+test(1913.17, key(sub.1), key(dt)) # reordering columns
+sub.2 <- subset(dt, a == 'a', select=c('a', 'c'))
+test(1913.18, key(sub.2), 'a')     # selected columns are prefix of key
+sub.3 <- subset(dt, a == 'a', select=c('b', 'c'))
+test(1913.19, is.null(key(sub.3))) # selected columns do not from a key prefix
+sub.4 <- subset(dt, a == 'cc')
+test(1913.21, nrow(sub.4), 0L)
+test(1913.22, is.null(key(sub.4)))
+#
+# transform maintains keys
+dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
+                 c=sample(20), key=c('a', 'b'))
+t1 <- transform(dt, d=c+4)
+test(1913.23, key(t1), key(dt))
+test(1913.24, t1$d, dt$c + 4)  # transform was successful
+t2 <- transform(dt, d=c+4, a=sample(c('x', 'y', 'z'), 20, replace=TRUE))
+test(1913.25, is.null(key(t2)))  # transforming a key column nukes the key
+## This is probably not necessary, but let's just check that transforming
+## a key column doesn't twist around the rows in the result.
+for (col in c('b', 'c')) {
+  test(1913.26, t2[[col]], dt[[col]])  # mutating-key-transform maintains other columns
+}
+#
+# tests-S4.R (S4 Compatability)
+#
+setClass("Data.Table", contains="data.table")
+setClass("S4Composition", representation(data="data.table"))
+# data.table can be a parent class"
+ids <- sample(letters[1:3], 10, replace=TRUE)
+scores <- rnorm(10)
+dt <- data.table(id=ids, score=scores)
+dt.s4 <- new("Data.Table", data.table(id=ids, score=scores))
+test(1914.1, isS4(dt.s4))
+test(1914.2, inherits(dt.s4, 'data.table'))
+## pull out data from S4 as.list, and compare to list from dt
+dt.s4.list <- dt.s4@.Data
+names(dt.s4.list) <- names(dt.s4)
+test(1914.3, dt.s4.list, as.list(dt))  # Underlying data not identical
+# simple S4 conversion-isms work
+df = data.frame(a=sample(letters, 10), b=1:10)
+dt = as.data.table(df)
+test(1914.4, identical(as(df, 'data.table'), dt))
+test(1914.5, identical(as(dt, 'data.frame'), df))
+# data.table can be used in an S4 slot
+dt <- data.table(a=sample(letters[1:3], 10, replace=TRUE), score=rnorm(10))
+dt.comp <- new("S4Composition", data=dt)
+test(1914.6, dt.comp@data, dt)
+# S4 methods dispatch properly on data.table slots"
+dt <- data.table(a=sample(letters[1:3], 10, replace=TRUE), score=rnorm(10))
+dt.comp <- new("S4Composition", data=dt)
+setGeneric("dtGet", function(x, what) standardGeneric("dtGet"))
+setMethod("dtGet", c(x="S4Composition", what="missing"), function(x, what){x@data})
+setMethod("dtGet", c(x="S4Composition", what="ANY"), function(x, what) {x@data[[what]]})
+test(1914.7, dtGet(dt.comp), dt)  # actually
+test(1914.8, identical(dtGet(dt.comp, 1), dt[[1]]))
+test(1914.9, identical(dtGet(dt.comp, 'b'), dt$b))
+# END port of old testthat tests
+
 
 ###################################
 #  Add new tests above this line  #
diff --git a/publish.R b/publish.R
index 48c6af31f3..d1182aedaf 100644
--- a/publish.R
+++ b/publish.R
@@ -123,7 +123,7 @@ check.copy <- function(job, repodir="bus/integration/cran"){
   dir.create(job.checks<-file.path(repodir, "web", "checks", pkg<-"data.table", job), recursive=TRUE);
   os = plat(job)
   from = file.path("bus", sprintf("%s/%s.Rcheck", job, pkg))
-  current.rout = c("main.Rout","main.Rout.fail","testthat.Rout","testthat.Rout.fail","knitr.Rout","knitr.Rout.fail","memtest.csv","memtest.png")
+  current.rout = c("main.Rout","main.Rout.fail","knitr.Rout","knitr.Rout.fail","memtest.csv","memtest.png")
   if (os=="Windows") {
     dir.create(file.path(job.checks, "tests_i386"), showWarnings=FALSE)
     dir.create(file.path(job.checks, "tests_x64"), showWarnings=FALSE)
@@ -159,7 +159,7 @@ check.index <- function(pkg, jobs, repodir="bus/integration/cran") {
     paste(na.omit(links), collapse=", ")
   }
   routs = lapply(jobs, function(job) {
-    current.rout = c("main.Rout.fail","testthat.Rout.fail","knitr.Rout.fail")
+    current.rout = c("main.Rout.fail","knitr.Rout.fail")
     os = plat(job)
     if (os=="Windows") {
       rout32 = file.path("tests_i386", current.rout)
diff --git a/tests/testthat.R b/tests/testthat.R
deleted file mode 100644
index c457c8fa5d..0000000000
--- a/tests/testthat.R
+++ /dev/null
@@ -1,6 +0,0 @@
-if(requireNamespace("testthat", quietly = TRUE)){
-    library(testthat)
-    library(data.table)
-    test_check("data.table")
-}
-
diff --git a/tests/testthat/test-S4.R b/tests/testthat/test-S4.R
deleted file mode 100644
index 13bc50744f..0000000000
--- a/tests/testthat/test-S4.R
+++ /dev/null
@@ -1,56 +0,0 @@
-context("S4 Compatability")
-
-## S4 class definitions to test
-setClass("Data.Table", contains="data.table")
-setClass("S4Composition", representation(data="data.table"))
-
-test_that("data.table can be a parent class", {
-  ids <- sample(letters[1:3], 10, replace=TRUE)
-  scores<- rnorm(10)
-  dt <- data.table(id=ids, score=scores)
-  dt.s4 <- new("Data.Table", data.table(id=ids, score=scores))
-
-  expect_true(isS4(dt.s4))
-  expect_true(inherits(dt.s4, 'data.table'))
-
-  ## pull out data from S4 as.list, and compare to list from dt
-  dt.s4.list <- dt.s4@.Data
-  names(dt.s4.list) <- names(dt.s4)
-  expect_identical(dt.s4.list, as.list(dt), info="Underlying data not identical")
-})
-
-test_that("simple S4 conversion-isms work", {
-  df = data.frame(a=sample(letters, 10), b=1:10)
-  dt = as.data.table(df)
-  expect_equal(as(df, 'data.table'), dt)
-  expect_identical(as(dt, 'data.frame'), df)
-})
-
-test_that("data.table can be used in an S4 slot", {
-  ## A class with a data.table slot
-  dt <- data.table(a=sample(letters[1:3], 10, replace=TRUE), score=rnorm(10))
-  dt.comp <- new("S4Composition", data=dt)
-  expect_equal(dt.comp@data, dt)
-})
-
-test_that("S4 methods dispatch properly on data.table slots", {
-  ## Make toy accessor functions and compare results against normal data.table
-  ## access
-  dt <- data.table(a=sample(letters[1:3], 10, replace=TRUE), score=rnorm(10))
-  dt.comp <- new("S4Composition", data=dt)
-  setGeneric("dtGet", function(x, what) standardGeneric("dtGet"))
-  setMethod("dtGet", c(x="S4Composition", what="missing"),
-            function(x, what) {
-              x@data
-            })
-  setMethod("dtGet", c(x="S4Composition", what="ANY"),
-            function(x, what) {
-              x@data[[what]]
-            })
-
-  expect_equal(dtGet(dt.comp), dt, label='actually')
-  expect_identical(dtGet(dt.comp, 1), dt[[1]])
-  expect_identical(dtGet(dt.comp, 'b'), dt$b)
-})
-
-
diff --git a/tests/testthat/test-data.frame-like.R b/tests/testthat/test-data.frame-like.R
deleted file mode 100644
index 8c7ec403bf..0000000000
--- a/tests/testthat/test-data.frame-like.R
+++ /dev/null
@@ -1,160 +0,0 @@
-context("data.frame like functions (merge, subset, transform)")
-
-###############################################################################
-## Merge
-test_that("`x` columns are valid (bug #1299)", {
-    d1 <- data.table(x=c(1,3,8), y1=rnorm(3), key="x")
-    d2 <- data.table(x=c(3,8,10), y2=rnorm(3), key="x")
-    ans1 <- merge(d1, d2, by="x")
-    ans2 <- cbind(d1[2:3], y2=d2[1:2]$y2)
-    setkey(ans2, x)
-    expect_equal(ans1, ans2, info="Original test #230")
-})
-
-test_that("`xkey` column names are valid in merge (bug#1299", {
-    d1 <- data.table(xkey=c(1,3,8), y1=rnorm(3), key="xkey")
-    d2 <- data.table(xkey=c(3,8,10), y2=rnorm(3), key="xkey")
-    ans2 <- cbind(d1[2:3], y2=d2[1:2]$y2)
-    setkey(ans2, xkey)
-    expect_equal(merge(d1, d2, by="xkey"), ans2, info="Original test #238")
-})
-
-test_that("one column merges work (bug #1241)", {
-    dt <- data.table(a=rep(1:2,each=3), b=1:6, key="a")
-    y <- data.table(a=c(0,1), bb=c(10,11), key="a")
-    expect_equal(merge(y, dt), data.table(a=1L, bb=11, b=1:3, key="a"),
-                 info="Original test #231")
-    expect_equal(merge(y, dt, all=TRUE),
-                 data.table(a=rep(c(0L,1L,2L),c(1,3,3)),
-                            bb=rep(c(10,11,NA_real_),c(1,3,3)),
-                            b=c(NA_integer_,1:6), key="a"),
-                 info="Original test #232")
-
-    ## y with only a key column
-    y <- data.table(a=c(0,1), key="a")
-    expect_equal(merge(y,dt), data.table(a=1L, b=1:3, key="a"),
-                 info="Original test #233")
-    expect_equal(merge(y, dt, all=TRUE),
-                 data.table(a=rep(c(0L,1L,2L),c(1,3,3)),
-                            b=c(NA_integer_,1:6),key="a"),
-                 info="Original test #234")
-})
-
-test_that("merging data.tables is almost like merging data.frames", {
-    d1 <- data.table(a=sample(letters, 10), b=sample(1:100, 10), key='a')
-    d2 <- data.table(a=d1$a, b=sample(1:50, 10), c=rnorm(10), key='a')
-    dtm <- merge(d1, d2, by='a', suffixes=c(".xx", ".yy"))
-    dtm.df <- as.data.frame(dtm)
-    dfm <- merge(as.data.frame(d1), as.data.frame(d2), by='a', suffixes=c('.xx', '.yy'))
-
-    expect_equal(unname(dtm.df), unname(dfm),
-                 info="Testing contents/data after merge")
-    expect_equal(colnames(dtm), colnames(dfm),
-                 info="Original test #255 (testing suffixes parameter)")
-})
-
-test_that("`suffixes` behavior can be toggled to pre 1.5.4 behavior", {
-    dt1 <- data.table(a=letters[1:5], b=1:5, key="a")
-    dt2 <- data.table(a=letters[3:8], b=1:6, key="a")
-
-    # options(datatable.pre.suffixes=FALSE)
-    # Option removed in 1.7.10
-    expect_equal(colnames(merge(dt1, dt2)), c("a", "b.x", "b.y"))
-
-    #options(datatable.pre.suffixes=TRUE)
-    #expect_equal(colnames(merge(dt1, dt2)), c("a", "b", "b.1"),
-    #             info="Pre 1.5.4 behavior not working")
-
-    #options(datatable.pre.suffixes=FALSE)
-})
-
-test_that("merge and auto-increment columns in y[x]", {
-  ## merging tables that have common column names that end in *.1 gets
-  ## tricky, because the y[x] mojo does some magic to increment the *.1
-  ## in the x (I think) and keep *.1 in the y
-  x <- data.table(a=letters[1:10], b=1:10, b.1=1:10 * 10, key="a")
-  y <- data.table(a=letters[1:10], b=letters[11:20], b.1=rnorm(10), key="a")
-  M <- merge(x, y)
-  m <- merge(as.data.frame(x), as.data.frame(y), by="a")
-
-  expect_is(M, 'data.table')
-  expect_is(m, 'data.frame')
-  expect_true(all(names(M) %in% union(names(M), names(m))))
-  for (name in names(m)) {
-    expect_equal(M[[name]], m[[name]])
-  }
-
-
-  ## Original example that smoked out the bug
-  M <- data.table(a=letters[1:10], b=1:10)
-  m <- as.data.frame(M)
-  ms <- lapply(1:3, function(x) data.table(a=letters[1:10], b=1:10 * 10^x))
-
-  for (i in 1:3) {
-    M <- merge(M, ms[[i]], by='a', suffixes=c("", sprintf(".%d", i)))
-  }
-
-  for (i in 1:3) {
-    m <- merge(m, as.data.frame(ms[[i]]), by='a',
-               suffixes=c("", sprintf(".%d", i)))
-  }
-
-  expect_is(M, 'data.table')
-  expect_is(m, 'data.frame')
-  expect_true(all(names(M) %in% union(names(M), names(m))))
-  for (name in names(m)) {
-    expect_equal(M[[name]], m[[name]])
-  }
-})
-
-###############################################################################
-## subset
-test_that("simple subset maintains keys", {
-  dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   c=sample(20), key='a')
-  sub <- subset(dt, a == 'b')
-  expect_equal(key(dt), key(sub))
-})
-
-test_that("subset using 'select' maintains key appropriately", {
-  dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   c=sample(20), key=c('a', 'b'))
-
-  sub.1 <- subset(dt, a == 'a', select=c('c', 'b', 'a'))
-  expect_equal(key(sub.1), key(dt), info="reordering columns")
-
-  sub.2 <- subset(dt, a == 'a', select=c('a', 'c'))
-  expect_equal(key(sub.2), 'a', info="selected columns are prefix of key")
-
-  sub.3 <- subset(dt, a == 'a', select=c('b', 'c'))
-  expect_true(is.null(key(sub.3)),
-              info="selected columns do not from a key prefix")
-
-  sub.4 <- subset(dt, a == 'cc')
-  expect_equal(nrow(sub.4), 0)
-  expect_true(is.null(key(sub.4)))
-})
-
-###############################################################################
-## transform
-test_that("transform maintains keys", {
-  dt <- data.table(a=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   b=sample(c('a', 'b', 'c'), 20, replace=TRUE),
-                   c=sample(20), key=c('a', 'b'))
-
-  t1 <- transform(dt, d=c+4)
-  expect_equal(key(t1), key(dt))
-  expect_equal(t1$d, dt$c + 4, info="transform was successful")
-
-  t2 <- transform(dt, d=c+4, a=sample(c('x', 'y', 'z'), 20, replace=TRUE))
-  expect_true(is.null(key(t2)), info="transforming a key column nukes the key")
-
-  ## This is probably not necessary, but let's just check that transforming
-  ## a key column doesn't twist around the rows in the result.
-  for (col in c('b', 'c')) {
-    msg <- sprintf("mutating-key-transform maintains other columns [%s]", col)
-    expect_equal(t2[[col]], dt[[col]], info=msg)
-  }
-})