We theorize and test the impact of two microfoundations of dynamic capabilities on incremental and radical innovation in knowledge-intensive businesses (KIBs). We argue that numerical flexibility and training, as human resources (HR) microfoundations that enable the configuration and deployment of resources, have different implications for incremental and radical innovation. Furthermore, we contend that those KIBs that make significant use of numerical flexibility will invest less in training, thus impacting innovation outcomes. We test our arguments using an original, longitudinal dyadic dataset assembled from 1750 structured interviews conducted with HR directors and senior managers in 875 KIBs located in Finland, France, Sweden and the UK. We find that numerical flexibility is negatively related to both incremental and radical innovation, whereas training is positively related to both. Our test of the moderating impact of training on both types of innovation resulting from the use of numerical flexibility was not statistically significant. Overall, we suggest that not all efforts by KIBs to mobilize resources and capabilities and align them with a complex external environment in the pursuit of innovation are necessarily beneficial. We highlight some research, policy, and practice implications that arise from our findings.