Did any of the values involved arise from a floating point computation? print can produce misleading output in such situations. For example:

$x = 1742-1e-12;
$y = 0;
$z = 1742;
print "$x + $y >= $z is ",
  ( $x + $y >= $z ? 'true' : 'false', "\n";
__END__
1742 + 0 >= 1742 is false
[download]

In this case, >= is correct; print is just misleading you. If this diagnosis is correct, the Numbers section of perlfaq4 is worth reading, as is perlnumber. To get trustier ints the general solution is to use sprintf. For example:

sub round {
  my $number    = shift;
  my $precision = shift || 0;  # optional precision
  my $template  = "%.${precision}f";
  return sprintf $template, $number;
}
$x = round( 1742-1e-12 );
$y = round( 0 );
$z = round( 1742 );
print "$x + $y >= $z is ",
  ( $x + $y >= $z ? 'true' : 'false', "\n";
__END__
1742 + 0 >= 1742 is true
[download]