- Google Trends  and the TIOBE index  show a steady decline over time 
- A wide reaching survey at Langpop places Perl near the bottom 
- A lot of Perl talk centers around people vigorously stating that Perl is not dead  and other kinds of cries for attention 
On the low end, PHP became the defacto web development language thanks to its ease of development and deployment. While many developers decry PHP's ugliness and its simplistic nature (no closures, no namespaces, 3000 functions), it nevertheless made web development far more accessible to everyone and thus allowed a huge number of people to make the products they wanted, and helped some notable companies (Flickr, Facebook etc).
On the other end, Python's steady improvement over the years (which made it both more powerful and easier than Perl) attracted a lot of companies and developers, while Ruby and Rails attracted developers who wanted more flexibility out of their language. While this was happening, there were no significant releases to Perl5 and Perl6 was nowhere in sight.
Another significant factor I believe was Perl's much beloved "There's More Than One Way To Do It" principle. While marketed to developers and giving them freedom and flexibility, it also hamper's Perls practicality.
First, it makes the language extraordinarily complicated. While most good designs struggle and strive for simplicity and coherence (the K.I.S.S. principle), Perl goes for the kitchen-sink mentality. Its table of operators can only be described by "gargantuan" and is only eclipsed by its grammar, which is at least 5 times larger than the grammars of languages like Python and Haskell. What this means is that Perl demands far more time out of a developer to become proficient at it, while offering no significant benefits for someone investing this extra time.
Second, it makes it hard for the community as a whole to progress. For example, while Python added new-style classes and pushed everyone into adopting them, there is no way for Moose to easily obtain the same ubiquity (as no one can push it and enforce it). In fact, it is rather telling that Perl's premiere web development framework, Catalyst, only acquired a decent object system in mid 2009.
Whether Perl will continue to decline or not is hard to say, though I highly doubt it'll ever return to its former glory - neither Perl5 nor Perl6 seem to offer much for developers who do not already love Perl.
Easter Egg: The Forex Chart Monkey gives advice on Picket Prevention and Forex Technical Analysis