I personally like analog scopes, but I do well more analog than digital stuff (probably 75% to 25%). BUT...and this is a big but, I've also got frequency counters, a logic analyzer, and various other doodads and whatsits that do a lot of the digital work for me.
Also, I really dont think 25 mhz is enough. You can ebay 60-100 mhz scopes for the same price, and if you go with an HP, tektronics, etc. it's going to work jsut fine unless it's seriously jacked up.
The bandwidth ratings are kinda different for digital and analog scopes. With an analog scope, it's generally the -3db bandwidth. This means that you'll get half the response at it's maximum bandwidth. You can actually use a 20mhz analog scope to look at a 40mhz signal, but you wont garner the information from that trace as you would if it were within the bounds of the scopes design. Pretty much, you'd get to look at a general representation of the waveform, but that about it. No accurate measurements at all.
With digital scopes, the bandwidth is generally indicating the response time of the ADC's inside. You cant, for instance, look at a 10 nanosecond pulse with a 25mhz digital scope even if it only repeats at 100hz (there are ways around this, but out of the scope of this post!). The rise times, jitter, etc of digital signals also fall into this category of inability.
For the most bang for your buck int he picaxe world, I'd have to suggest a digital scope. However, I'd also suggest something much higher than 25mhz. I'd look at 50mhz at least. 100mhz would be preferable. It just gives you a ton more information than a lower bandwidth scope would give you and as a tool is much more versatile in the long run for many activities.
--Andy P