We report on active control of the femtosecond photodissociation and ionization reactions of iron pentacarbonyl, Fe(CO)(5), in the gas phase. The spectral phase of femtosecond laser pulses is modified in a pulse shaper, employing a learning evolutionary algorithm, Direct feedback from the experiment is used to iteratively improve the laser pulse shape according to a given optimization problem. This many-parameter optimization is compared with one-parameter control schemes and found to be more versatile, because it can sample a much more general search space. Information about the underlying reaction mechanism can be extracted from the results of the automated optimization. It is further shown that second-harmonic generation (SHG) can be used at the output of a 800 nm pulse shaper to implement 400 nm excitation experiments. The optimization procedure not simply increases the SHG efficiency but optimizes the objective given for the combined system of SHG and Fe(CO)(5) photochemistry. The importance of the choice of fitness function is examined experimentally. By choosing appropriate weighting factors, it is possible to tune the optimization results from an optimization of the ratio of photoproduct yields toward optimization of the absolute photoproduct yields. Evolutionary laser pulse shaping is considered a very useful tool for unraveling the processes of photoinduced chemical reactions.