Both scientists and farmers are confronted by a similar question: which current and past cropping system components will influence the present weed flora, and how? This information is necessary to optimize both cropping systems for weed control, and quality and cost in surveys and monitoring schemes. The present study addressed these questions with a sensitivity analysis to input variables of a cropping system model, AlomySys, that predicts weed dynamics in interaction with pedo-climatic conditions. The study ranked cropping system components according to their impact on weed infestation in winter wheat, showing for instance that though crop succession was crucial, current and past tillage strategies influenced grass weed densities even more. Crops were not only ranked as a function of the resulting weed risk but the latter was also linked to crop species traits, i.e. crop type, usual sowing period and emergence speed. A previous winter v. spring crop thus increased weed density by 72% in the following winter wheat; a late-sown v. early sown winter crop by 26%, a slow v. fast-emerging winter crop by 17%, and a lower competitive ability by 9%. Similarly, the characteristics of each crop management technique (tillage, catch crop, secondary crop, mowing, mechanical weeding, herbicides, nitrogen fertilizer, manure and harvest) were quantified. For instance, the timing of the first tillage operation was crucial prior to the analysed winter wheat crop while the choice of the tool used even 5 years previously still influenced weed infestation in the current year; a catch crop prior to previous spring sown crops reduced the current infestation regardless of catch crop sowing dates and densities, but the reductive effect could be lost if the field was tilled several times to destroy the catch crop. The advice synthesized here and in a companion paper (Colbach & Mézière 2012). will be valuable to design innovative, integrated cropping systems, indicating (1) which cropping system components to modify to produce the largest effect, (2) for how long past practices must be considered when choosing current options and (3) the optimal options for the different management techniques. Points (1) and (2) are also valuable to identify data to record in surveys, though still resulting in a total of 232 variables. In a second step, these detailed variables were therefore simplified and aggregated to determine a smaller set of 22 synthetic variables easily recorded in surveys, such as the proportion of winter and spring crops during the last 10 years (instead of the actual crop sequence), the proportion of crops sown in summer, early autumn, late autumn, early spring and late spring during the last 5 years (instead of exact sowing dates), the ploughing frequency (instead of ploughing dates and characteristics), the mean number of herbicide sprayings per year (instead of dates), etc. This reduced survey list will reduce the cost of surveys as well as increase the number and quality of surveys as more farmers will be ready to participate and there will be fewer uncertainties in the answers.