Don't be afraid to use the Perl regex engine in a couple of lines! The speed of this thing in Perl 5.10 vs Perl 5.6 is stunning. GO FOR CLARITY first.

So talking thru the code below: (1) process only things that are US dollars (optional), (2) there is gonna be zero or more spaces, followed by the "$" sign and some sequence of digits,commas and periods. (3)that sequence gets captured and then the commas get deleted.

Update: If max valid dollar amount is $100,000.00 then I would enforce the 100,000 max via a numeric comparison on $number. BUT, this is actually the wrong place in the system to be checking that!

#!/usr/bin/perl -w
use strict;

my @bucks = ('$1,999,003 USD ', '$1,532.33 AUS', '$1,500.11 USD');

foreach my $amount (@bucks)
{
   next unless $amount =~ m/USD/;
   my ($comma_amount) = ($amount =~ /\s*\$([\d,.]+)/);
   print "c=$comma_amount\n";
   (my $number = $comma_amount) =~ s/,//g;
   print "n=$number\n";  
}
__END__
prints:
c=1,999,003
n=1999003
c=1,500.11
n=1500.11
[download]

/^\$(?:100,?000(?:\.00)?|\d{1,2},?\d{3}(?:\.\d\d)?)$/
[download]

#!/usr/bin/perl --

use strict;
use warnings;

use Regexp::Assemble;

my( @usd ) = (
    q[$0,000.00],
    q[$0000],
    q[$999],
    q[$1,000],
    q[$1,000.00],
    q[$2,000],
    q[$9,000],
    q[$10,000],
    q[$10,00000.00],
    q[$100,000],
    q[$999,000],
    q[$1,000,000],
    q[$1,234,567,890.00],
);

sub refit {
    my( $re ) = Regexp::Assemble->new;
    $re->add(shift) while @_;
    $re=$re->re;
    print "\n\n$re\n";
    for my $usd ( @usd ){
        no warnings 'uninitialized';
        printf "[ %20s ]= %d\n", $usd, $usd =~ $re;
    }
    print '-' x 50 , "\n\n";
}

refit(
    '^\$1,?\d\d\d(?:\.\d\d)?$',   # $1,000  $1,000.00
    '^\$[2-9],?\d\d\d(?:\.\d\d)?$',   # $2,000 $9,000
    '^\$\d\d,?\d\d\d(?:\.\d\d)?$',   # $10,000 $99,000
    '^\$\d\d\d,?\d\d\d(?:\.\d\d)?$',   # $100,000  $999,000 
    '^\$\d,?(?:\d\d\d,?)+\d\d\d(?:\.\d\d)?$',   # $1,000,000  $1,234,5
+67,890.00
);

refit(
    '^\$1,?\d\d\d(?:\.\d\d)?$',   # $1,000  $1,000.00
    '^\$[2-9],?\d\d\d(?:\.\d\d)?$',   # $2,000 $9,000
    '^\$\d\d[\d,]*?\d\d\d(?:\.\d\d)?$',   # $10,000 $10,00000.00
);

__END__




(?-xism:^\$(?:\d(?:,?(?:\d\d\d,?)+|\d?\d,?)|(?:[2-9]|1),?)\d\d\d(?:\.\
+d\d)?$)
[            $0,000.00 ]= 0
[                $0000 ]= 0
[                 $999 ]= 0
[               $1,000 ]= 1
[            $1,000.00 ]= 1
[               $2,000 ]= 1
[               $9,000 ]= 1
[              $10,000 ]= 1
[         $10,00000.00 ]= 0
[             $100,000 ]= 1
[             $999,000 ]= 1
[           $1,000,000 ]= 1
[    $1,234,567,890.00 ]= 1
--------------------------------------------------



(?-xism:^\$(?:(?:[2-9]|1),?|\d\d[\d,]*?)\d\d\d(?:\.\d\d)?$)
[            $0,000.00 ]= 0
[                $0000 ]= 0
[                 $999 ]= 0
[               $1,000 ]= 1
[            $1,000.00 ]= 1
[               $2,000 ]= 1
[               $9,000 ]= 1
[              $10,000 ]= 1
[         $10,00000.00 ]= 1
[             $100,000 ]= 1
[             $999,000 ]= 1
[           $1,000,000 ]= 0
[    $1,234,567,890.00 ]= 0
--------------------------------------------------
[download]

Use the regex engine just to locate the number. To see if the number is between a thousand and a hundred thousand, use the built-in Perl implementation of arithmetic.

while (<>) {
    chomp;
    print "$_: ";        
    if (my ($number) = /^\$?(\d[,\d]*)/) {
        $number =~ s/,//g;
        print $number < 1000    ? 'too small' :
              $number > 100_000 ? 'too big'   :
                                  'just right';
    }
    else {print 'not a number'}
    print "!\n";
}
[download]

$re = qr!
    ^
    \$\d
    [\d,]*?
    \d\d\d
    (?:\.\d\d)?
    $
    !x;
[download]

^\$(?:(?:1\d{0,2})|(?:[1-9]\d?)),\d{3}(?:\.\d\d)?$
[download]