When reading the file you normally skip the header line. So spaces in the variable names shouldn't matter. data want ; infile 'myfile.dat' firstobs=2 truncover ; length name $20 Marks Age 8 DOBchar $12 ; input name -- DOBchar ; run; If you are using PROC IMPORT or some other tool that will try to guess at what your file contains then it will have algorithms for converting the invalid characters into valid variable names.   As to converting the gibberish in the DOB field you probably will need to read it as character and write your own algorithm to see if you can convert it to a real date. ... View more