Since the "surprise" eruption of Chaiten in southern Chile (still erupting away), I'm sure there has been a lot of talk about better monitoring and predictions for volcanic eruptions. Now, we don't know the full extent of the facts, but usually an eruption of the magnitude of Chaiten (VEI 6, i.e., BIG) don't just go off out of the blue. There are precursors, such as seismicity under the volcano, uplift of the land over the volcano (think of the bulge of Mt. Saint Helens prior to the 1980 eruption), increasing emissions of volcanic gases such as carbon dioxide or sulfur dioxide, increase in stream/spring water temperatures near the volcano, increase in the temperature of the land at the volcano. These are just some of the ways to watch volcanoes looking for signs of potential eruptions.
However, it is not financially possible to wire up every volcano in a volcanically active country like Chile (or the U.S. for that matter) to take every vital statistic of the system. What you need to do is think about probability and hazard. A volcano with a history of eruptions near population ranks high on the list, while a remote volcano with little record of activity (although not extinct) is low. Now, there will always be ringers like Chaiten, whose recent eruptive history would have put in lower on such a list, but looking at the type of activity, whether it is large or small eruptions and their recurrence interval, can help in making some volcanoes that might not be perceived as dangerous get the attention they deserve.
In order to do this, you need a number of things. Firstly, you need to know your volcanoes. This requires field work and active research of the history and style of eruption of particular volcanoes (hint hint NSF). Secondly, you need people and equipment to monitor the volcano - at the least in a passive, "look for the warning signs" way. Thirdly, you need people who can interpret the symptoms when things begin to change. Building such a system from the ground up can be a daunting task for many developing nations in light of the expertise and technology needed to do it right.
Which brings me to the USGS press release from today, announcing that the USGS will be helping Chile construct an early warning system for its volcanoes. This is not to say that Chile wasn't already monitoring their 120 volcanoes, but this new system will be modeled after the US system that ranks volcanoes by their potential activity and hazard. It will also help the Chileans construct probability trees for the potential activity at a volcano when it does start to show signs of eruption. The USGS has a good track record for helping countries build effective monitoring networks - see Colombia and the Philippines. It is too bad that a large eruption has to occur to get people thinking about the dangers of active volcanoes.