哇，还没碰到过这个数量级的dataset

AFAIK there is no limitation as long as your system hardware can handle in a data step.   Try no using import for text file. The example below reads and writes 10**8 rows.

12   data _null_;
13      file 'c:\temp\i.txt';
14      do i=1 to 1e8;
15        put i;
16      end;
17   run;

NOTE: The file 'c:\temp\i.txt' is:
      Filename=c:\temp\i.txt,
      RECFM=V,LRECL=256,File Size (bytes)=0,
      Last Modified=18Jan2013:18:17:29,
      Create Time=18Jan2013:18:16:39

NOTE: 100000000 records were written to the file 'c:\temp\i.txt'.
      The minimum record length was 1.
      The maximum record length was 9.
NOTE: DATA statement used (Total process time):
      real time           26.13 seconds
      cpu time            18.70 seconds


18
19   data _null_;
20      infile 'c:\temp\i.txt';
21      input i;
22   run;

NOTE: The infile 'c:\temp\i.txt' is:
      Filename=c:\temp\i.txt,
      RECFM=V,LRECL=256,File Size (bytes)=988888898,
      Last Modified=18Jan2013:18:17:55,
      Create Time=18Jan2013:18:16:39

NOTE: 100000000 records were read from the infile 'c:\temp\i.txt'.
      The minimum record length was 1.
      The maximum record length was 9.
NOTE: DATA statement used (Total process time):
      real time           17.26 seconds
      cpu time            17.08 seconds

谢谢，后来我用了infile，还是只能录入700多万。而且sas不报错。最后BI同事说，是因为原来文件从DB2导出 的，DB2有自己大型文件的格式，导入时候会无法识别。所以后来就找到后面2百万数据里面有点乱码，去除掉，就导入了。楼上兄弟对的，sas没有这个限制。谢谢哈

可不可以先import前4m个，在输入后面的

are u a girl?