Pions are the least massive of all mesons, and thus are also the lightest hadrons. There are three types of pion; the positive pion (π+), the negative pion (π-), and the neutral pion (π0). Pions are composed solely of up quarks and down quarks , and have spin 0. The pions have a mass of about 140 MeV/c^2, or about 3/20 of that of a proton.
Pions, being mesons, are pairs of one quark and one antiquark. Up quarks have charge +2/3 and down quarks have charge -1/3, and their respective antiparticles have the opposite charge, so there is only one combination that gives a positive pion: an up plus an antidown. I will write this as ud*, although this is not the usual convention. Similarly, there is only one combination that gives a negative pion: du*. This makes the π- the antiparticle of the π+. For the π0, however, things are more complicated. There are two neutral combinations available: uu* and dd*. So are there then two different π0s?
The answer, somewhat strangely, is no. The actual, observed π0 is a superposition of states; its form is 1/sqrt(2)(uu* - dd*). This means that a real π0 is half of a uu* combined with half of a dd* in a particular way. (1/sqrt(2)(uu* + dd*) is a different, spin 1 particle called the omega meson) This is actually an example of a fairly common occurrence in particle physics, where an observed particle is the superposition of multiple different sets of constituents. This produces more observable effects in the next largest class of meson, the kaon (see that node for details). Also, it is possible for multiple particles to have the same set of constituents; the rho mesons have the same constituents as the pions but have spin 1.
Notice that all of the constituents of the π0 consist of a quark and its corresponding antiquark. Since particles annihilate with their corresponding antiparticles, how can the π0 exist as a particle of its own? At the time scales studied by particle physics, nothing happens 'instantly', but rather has a rate or a lifetime associated with it. So a meson consisting of a quark and an antiquark of the same flavour will not immediately self-annihilate. Nevertheless, it will not be very stable and indeed the lifetime of the π0 is dramatically shorter than that of the π+ and π-.
Pions are all unstable particles and they decay a short time after their production. Charged pions have an average lifetime of approximately 25 nanoseconds and the π0 has an average lifetime of only 0.08 femtoseconds, a factor of one billion less than that of the charged pions. When charged pions decay, they usually (99.998% of the time) decay into a muon of the same charge and a muon neutrino. π0s generally decay into pairs of gamma ray photons, although about 1% of the time they decay into one gamma ray and an electron/positron pair.
Pion physics is currently a fairly well understood part of particle physics; most of the interesting meson experiments going on today study kaons or B mesons. Nevertheless, it was once an active area of experimentation, with several particle accelerator facilities built in the 1970s as 'pion factories', such as the Canadian accelerator lab TRIUMF. The understanding of pion physics gained by these experiments is useful in processing newer, higher-energy particle physics experiments since most high-energy particle reactions produce pions at some stage of the process. Pions are also crucial to the binding of protons and neutrons in nuclei, but someone more well-versed in nuclear physics than I will be needed to explain this.
Sources include my senior undergraduate particle physics course and the (very technical) Particle Data Group website at http://pdg.lbl.gov/
This writeup is copyright 2004 D.G. Roberge and is released under the Creative Commons Attribution-NoDerivs-NonCommercial licence. Details can be found at http://creativecommons.org/licenses/by-nd-nc/2.0/ .