•Differentiated knowledge bases are measured using linked-employer-employee micro data.•Analytical, synthetic, and symbolic knowledge has a positive impact on firm growth.•Growth-enhancing effects of combinations of the three knowledge bases are detected.•The relationship between knowledge bases intensities and firm growth is curvilinear.•The results are robust in a wide range of models. The link between knowledge and firm growth has been a core topic in economics of innovation for a long time. However, despite strong theoretical arguments, empirical evidence remains inconclusive. One important reason for this conundrum may be the failure of standard indicators to capture firm innovation activities comprehensively. We contribute to overcoming this limitation by looking in the knowledge processes that drive variegated forms of innovation and aim thereby to establish a solid relationship with firm growth in more detail. Our arguments draw on the differentiated knowledge base approach, distinguishing between analytical, synthetic, and symbolic knowledge. We measure the three types of knowledge bases with detailed longitudinal linked-employer-employee micro-data from Sweden. Econometric findings based on a very large sample of small and medium-sized firms indicate significantly positive effects of the three knowledge types, and in particular combinations thereof, on firm growth. In addition, we show that not only high-growth but also slow-growth firms benefit immensely from the use of combinatory knowledge bases. We find evidence on a curvilinear relation between knowledge bases and growth of firms. Beyond certain thresholds increasing the knowledge bases further results in decreasing firm growth. Our results remain robust in a wide range of specifications and econometric models.