haven write_sas() in R: Export Data Frames to SAS Files

The haven write_sas() function exports an R data frame to a SAS .sas7bdat file in a single line. Because the SAS format is proprietary, write_xpt() is the more reliable choice when SAS itself must open the file.

library(haven)                           # load the package
write_sas(df, "data.sas7bdat")           # write a .sas7bdat file
write_xpt(df, "data.xpt")                # SAS transport file (reliable)
write_xpt(df, "data.xpt", version = 5)   # legacy SAS 5 transport
read_sas("data.sas7bdat")                # read the file back in
write_dta(df, "data.dta")                # Stata instead of SAS

What write_sas() does

write_sas() writes an R data frame to a SAS data file. It belongs to the haven package, which connects R to the binary formats used by SAS, Stata, and SPSS. You pass a data frame and a destination path ending in .sas7bdat, and haven serializes the data, column types, and any variable labels into one SAS-native file. Haven builds on the ReadStat C library, so the conversion runs without a SAS installation on your machine.

write_sas() syntax

The function takes just two arguments. The signature is write_sas(data, path). The data argument is the data frame or tibble you want to export, and path is the file path where the .sas7bdat file is written. write_sas() returns the input data invisibly, so you can drop it into a pipe without breaking the chain.

Column types map to SAS predictably. Numeric and integer columns become SAS numeric variables, character columns become SAS character variables, and labelled columns carry their value labels across. Factors are converted to character on export, so the levels survive as text but the ordering attribute is lost.

Export a data frame to a SAS file

A single call writes the file to disk. Start by loading haven and building a small data frame to export.

Pass that data frame and a path to write_sas(). Writing to tempdir() keeps the example self-contained and avoids cluttering your working directory.

The file now exists on disk. Read it back with read_sas() to confirm the round trip preserved every column.

Why SAS may not open your file

This is the most important caveat with write_sas(). The .sas7bdat format is proprietary and undocumented, so haven reverse-engineers it. Files that write_sas() produces are frequently rejected by SAS itself, even though R and haven read them back without trouble. If the file only travels between R sessions or other ReadStat-based tools, .sas7bdat is fine. If a SAS user must open it, use write_xpt() instead.

write_xpt() writes a SAS transport file (.xpt), a format that SAS publicly documents and supports through PROC CIMPORT and the XPORT engine.

The version argument controls compatibility. Setting version = 8 (the default) allows 32-character variable names, while version = 5 matches older SAS releases but truncates names to 8 characters.

Common pitfalls

Three mistakes account for most failed SAS exports.

1. SAS rejects the .sas7bdat file. The native format is unreliable from R. Switch to write_xpt() whenever the file leaves the R ecosystem.
2. Variable names get truncated. SAS variable names cap at 32 characters, and version = 5 transport files cap at 8. Long, descriptive R names are silently shortened.
3. Unsupported types do not survive. R NA becomes a SAS missing value correctly, but list columns and some date-time classes coerce unexpectedly. Convert them to plain Date or numeric columns first.

The truncation pitfall is easy to trigger and easy to miss.

Try it yourself

Related haven functions

haven exports to every major statistical format. Once write_sas() is familiar, these siblings cover the rest of the workflow.

• write_xpt() writes SAS transport (.xpt) files that SAS reads reliably.
• read_sas() reads .sas7bdat files back into R as a tibble.
• write_dta() exports a data frame to a Stata .dta file.
• write_sav() exports a data frame to an SPSS .sav file.
• read_xpt() reads SAS transport files into R.

For the full argument reference, see the official write_sas() documentation.

FAQ

Can SAS open files created by write_sas()?

Often it cannot. The .sas7bdat format is proprietary and undocumented, so haven reverse-engineers the layout. SAS frequently rejects files written this way, even though R reads them back fine. When a SAS user must open your data, export with write_xpt() instead, which produces a documented transport file that SAS reads reliably through its XPORT engine.

What is the difference between write_sas() and write_xpt()?

write_sas() writes a native .sas7bdat file, while write_xpt() writes a .xpt transport file. The transport format is publicly documented, so SAS opens it consistently. The native format is not documented, so haven's best-effort writer often produces files SAS cannot read. Use write_xpt() for handoff to SAS and write_sas() only when the consumer is R.

How do I export an R data frame to SAS?

Load haven, then call write_sas(df, "path.sas7bdat") for a native file or write_xpt(df, "path.xpt") for a transport file. Both take a data frame and a path. For real SAS compatibility, prefer write_xpt(). The function creates the file on disk and returns the data frame invisibly so it can be used inside a pipe.

Does write_sas() preserve variable labels?

Yes. Columns created with haven's labelled class carry both their variable labels and value labels into the SAS file. Plain numeric and character columns export without labels. Factors are converted to character, so the level text survives but the ordered attribute does not. Round-trip the file with read_sas() to confirm the metadata you expect came through.

What file extension does write_sas() use?

write_sas() expects a path ending in .sas7bdat, the extension for native SAS data sets. write_xpt() uses .xpt for transport files. Using the correct extension matters because downstream tools and SAS engines select their reader based on it. A mismatched extension can leave a file that no program recognizes.