We present an ab initio study of gaseous SO2 and O-3(-)(H2O)(n) collisions. Opposed to the usual approach to determine reaction rates via structural optimizations and transition state theory, we successfully approach this problem using ab initio molecular dynamics. We demonstrate the advantages of this approach, being the automatic and unbiased inclusion of dynamic and steric effects as well as the simultaneous assessment of all possible reactions. For this particular system, we find that only one reaction will be of atmospheric significance. Further, we identify the main geometrical parameters governing and limiting the observed reaction and suggest a new measure of the reaction rate being ca. 3/4 of the collision rate.