It is widely recognized that mineral fertilizers must play an important part in improving agricultural productivity in western Kenyan farming systems. This paper suggests that for this goal to be realized, farmers??? knowledge must be strengthened to improve their understanding of fertilizers and their use. We analyzed smallholder knowledge of fertilizers and nutrient management, and draw practical lessons from empirical collective fertilizer-response experiments. Data were gathered from the collective fertilizer-response trials, through focus group discussions, by participant observation, and via in-depth interviews representing 40 households. The collective trials showed that the application of nitrogen (N) or phosphorous (P) alone was insufficient to enhance yields in the study area. The response to P on the trial plots was mainly influenced by incidences of the parasitic Striga weed, by spatial variability or gradients in soil fertility of the experimental plots, and by interactions with N levels. These results inspired farmers to design and conduct experiments to compare crop performance with and without fertilizer, and between types of fertilizers, or responses on different soils. Participating farmers were able to differentiate types of fertilizer, and understood rates of application and the roles of respective fertilizers in nutrient supply. However, notions were broadly generated by unsteady yield responses when fertilizers were used across different fertility gradients, association with high cost (especially if recommended rates were to be applied), association of fertilizer use with hybrids and certain crops, historical factors, among other main aspects. We identified that strengthening fertilizer knowledge must be tailored within existing, albeit imperfect, systems of crop and animal husbandry. Farmers??? perceptions cannot be changed by promoting more fertilizer use alone, but may require a more basic approach that, for example, encourages farmer experimentation and practices to enhance soil properties such as carbon build-up in impoverished local soils.