gfortran 4.2 and 4.3 (but not GNU Fortran 95 (GCC) 4.1.2 (Gentoo 4.1.2 p1.1)) violate the standard with a format of the form 4P, F14.0. Reproducible: Always Steps to Reproduce: Compile and run print '(f14.0,4pf14.0 )', 3742.0, 0.3742 end Actual Results: 3742. 0000. Expected Results: 3742. 3742. According to a long time member of X3J3 I agree that the discussion of P editing in 10.6.5.1 is silent concerning F editing on output, which might lead one to conclude that it has no effect. But the last paragraph about F editing in 10.5.1.2.1 (167:23) says "modified by the established scale factor." So I think the new compiler is broken. Four of my compilers (lf95, NAG, Intel, g95 (4.0.3 0.91!)) give your old results. gfortran (4.1.2 20071124) gives your new results, which I think are wrong.
we really dont track fortran standards compliance ... you should be filing a bug here: http://gcc.gnu.org/bugzilla/