We describe how active feedback routines can be applied at limited repetition rate (5 Hz) to optimize highpower (> 10 TW) laser interactions with clustered gases. Optimization of x-ray production from an argon cluster jet, using a genetic algorithm, approximately doubled the measured energy through temporal modification of the 150 mJ driving laser pulse. This approach achieved an increased radiation yield through exploration of a multi-dimensional parameter space, without requiring detailed a priori knowledge of the complex cluster dynamics. The optimized laser pulses exhibited a slow rising edge to the intensity profile, which enhanced the laser energy coupling into the cluster medium, compared to the optimally compressed FWHM pulse (40 fs). Our work suggests that this technique can be more widely utilized for control of intense pulsed secondary radiation from petawatt-class laser systems.Petawatt lasers are now able to operate with pulse repetition rates of 1 Hz 1 and upcoming facilities using more efficient, lower thermal load diode-pumped solid state technology will increase this to 10 Hz 2 or more. One of the major drivers for the increase in average power of such high peak-power systems is to generate bright laserplasma based secondary sources to provide user beamlines, similar to existing light-source facilities, or energetic particle beams for a range of applications. This move to a higher repetition rate opens the possibility to employ active feedback routines to optimize energy conversion into radiation or particle beams. Due to highly complex non-linear dynamics in intense laser-plasma interactions, the optimal laser pulse properties for generation of the secondary source can not easily be predicted, as this is both computationally demanding and requires a complete understanding of all the key physical processes involved. Also, optimization is a many-dimensional problem and so cannot readily be performed by scanning individual parameters.Sophisticated feedback techniques, usually employing kHz repetition rate lasers operating at relatively low peak intensity, are well-established for coherent control of atomic and molecular processes 3 . Programmable elements in the laser system are used to tailor the spatial and temporal profile of the laser pulse at focus to optimize specific output parameters. One method is to use a genetic algorithm (GA) to select the most suitable profiles out of an initially random or pseudo-random set and, over a number of generations, the input parameters are evolved to improve the chosen output property (referred to as the fitness function). The great benefit of this approach is that it can achieve advantageous results without detailed knowledge of the interaction itself, and lead to new and unexpected results.Previous experiments have employed feedback loops to control high harmonic generation 4,5 , cluster dynamics 6-8 and electron beam properties 9 through temporal and spatial pulse shaping. These studies were performed with low energy pulses (< 20 mJ) and, with the excepti...