Is there any thing other than time that "triggers" a radioactive atom to decay? Say you have a vial of tritium and monitor their atomic decay with a geiger counter. How does an atom "know" when it's time to decay? It seems odd that all the tritium atoms are identical except with respect to their time of decay.
 A: You can trigger decay of certain nuclei with gamma rays, just like you can stimulate emission of photons from excited atoms with incoming radiation. You can even make a bomb if that is your kind of thing. Induced emission
A: Actually, all the atoms are identical. The time at which it is observed to decay is not an intrinsic property of a given atom, but rather an effect of quantum mechanics. For any given time bin, there is some finite amplitude for a transition to a decayed state, which in turn corresponds to a finite probability, since the particle(s) emitted will escape from the system once such a state is reached. This also means that the process is in irreversible, due to the open nature of the system. This works in the same way as atomic transitions when atoms emit photons (see the relevant Wikipedia page).
For each undecayed atom, in each time bin $T$ there is a probability of transitioning to the decayed state given by a fixed probability $p$ (which is independent of $T$, and depends only on the binning size). Thus between the time $t$ and $t+\Delta t$ there is a fixed probability $\Delta p = \lambda \Delta t$ of transitioning to the decayed state for any given atom. So if we have $N(t)$ undecayed nuclei at time $t$, then at time $t+\Delta t$ we should have $N(t+\Delta t) = (1-\lambda\Delta t)N(t)$. Rearranging thisa and taking the limit $\Delta t \to 0$ we obtain $dN/dt = -\lambda t$. Solving this equation yields the total number of nuclei left undecayed at time $t$ as $N(t) = N(0) e^{-\lambda t}$.
Anyway, the point to take from all this is simply that the atoms are all identical and decay by a purely random process.
UPDATE: I forgot to mention that decay probability can be increased, for example via collision with another particle for the right energy, and this is exactly how fission based nuclear bombs work. Here though, again, there is nothing special about the particular atom decaying, and it is simply the particles involved in the collision that have the increased decay probability. (I must admit that I have pared this picture right down to the basics as otherwise it would need to be a far more technical discussion).
A: I think if even there were a "trigger variable" for each atom, one would need to randomize it anyway to describe an ensemble of decaying atoms.
On the other hand, in case of atoms there is a stimulated emission - with help of photons coherent with the "future" photon. This shows that the "environment" is somewhat important. As soon as the environment is complicated and is hard to control, one can loosely think that the random character of decays is due to random character of the "triggering QM environment".
A: There are more reasons then simple random time as we can see in this image.
 (M. Yamamoto et. al. Journal of Environmental Radioactivity, 2006, 86, 110-131)  (from WP )
EDIT add 1
This data is based in sediments. As sediments trace the environment conditions, they act as a proxy.  
The next image is a variation of The decay rate of the radioactive isotope 32Si, and is found in
 "“Evidence for Correlations Between Nuclear Decay Rates and Earth-Sun Distance”  by Jenkins et al. 2008.
Later the distance from Sun connection was dismissed, but data is still valid. 
Similar data about neutrino and Wimp seasonal variarance can be found.
EDIT add 1 end
EDIT add 2
a related paper, in the opposite direction can be found here
"Evidence against correlations between nuclear decay rates and Earth-Sun distance"
"We have reexamined our previously published data .. We find no evidence for such correlations"
They used ratios and this null result is expectable if both the samples and the 'presumed' reference are affected by the same nuclear processes. IMO, in this situation a counting procedure is better than to use ratios.   

If the Jenkins proposal were correct,
  it is very unlikely that the alpha,
  beta-minus, beta-plus, and
  electron-capture decays of all
  radioactive isotopes would be affected
  in quantitatively the same way. Thus
  the ratios of counts observed from two
  different isotopes would also be
  expected to show annual variations.
In order to minimize the influence of
  any changes in detector and/or
  electronics performance, we analyzed
  ratios of gamma-ray peak areas from
  the isotope of interest and those from
  a reference isotope whose half life
  was well known.

EDIT add 2 end
There are seasonal variations (diurnal and annual) in radioactive processes:


*

*Radon, Lead, etc ...  

*Neutron production on reactors 
(at lab   and in space missions
RTGs)

*Neutrino

*DM - WIMPs 


I dont know what the experts say about the actual explanations. I think that they dont know the whys. 
I'm in the chase of data labeled with the timedate and geographical local of the 'crime'.
Can someone help, pls? 
Are the atoms of the same isotope equal?
I can not agree with the most voted answer, by Joe:

"Actually, all the atoms are
  identical. The time at which it is
  observed to decay is not an intrinsic
  property of a given atom, but rather
  an effect of quantum mechanics."

Since when quantum mechanics has effects? QM does not produce any effect, QM describes what we see at a statistical level.
We see that at a particular time moment, a particular atom decayed, and not the other. It  has to be an intrinsic property of that particular atom that made it decay at that precise moment.  
I do not know of any experiment that tried to measure how much equal or distinct can be a similar group of atoms. The community has  the hope that they are identical. 
I'm skeptical about this issue and I take nothing from granted, that I can say 'They are different one from the others'.  
A: It seems to be widely assumed that the observed click in the geiger counter corresponds to the instantaneous decay of a particular tritium atom. I don't know if I'm just pointing out the obvious, but I'm quite sure this correspondence has never been explicitly demonstrated. Quantum mechanics tells us there is a certain flux of electrons emanating from the vial of tritium; that there are is a certain frequency of clicks in a geiger counter; and that if analyzed, the sample of tritium may be separated into two streams, one of which turns out to be helium. These are three different phenomena, none of which can be easily correlated with any of the others. 
To put it plainly, all we can say about your sample of tritium is that the atoms are in a superposition of states. When they are observed individually, they are found to be in one or the other atomic state - tritium or He3. There is no experiment I know of where we can identify the moment when a particular tritium atom changed state.
A: Sir Isaac Newton struggled with exactly this question, in the context of optics, and the best he could come up with was a theory of superluminal waves associated with light particles.
(The question Newton asked was essentially the same question as yours: If a light beam is 20% absorbed into a glass surface and 80% reflected, and if light is particular, then how can one particle acting alone make a correct decision?)
The whole business of selecting one from a possible set for no apparent reason can be used for quantum computing. It can effectively compute a "diagonalisation of a matrix". 
Lets look more closely at the meaning of random.
Random is the limit of compressibility of a pattern, the removal of all predictability.
The pattern as a whole has this character, so it is a feature of the set.
A radioactive particle does not have the knowledge of which you speak, in fact it is missing information. It is forbidden to carry predictive knowledge. No information can be projected from any part of the sequence, present or future, to determine any part of the sequence. In this sense, every event in the sequence has no extractable knowledge about its position in the sequence. It is a safe spy, unable to betray its fellows.
This does not even answer the question, but gives a better way to think about it. What we see about randomness as contrived should more be seen as a natural default. Einsten protests that God does not play dice, but he is possibly complaining if God does plays dice then that means that God will always be with-holding information. 
