Original equipment manufacturers increasingly involve suppliers in new product development (NPD) projects. How companies design a contract to motivate supplier participation is an important but under-examined empirical question. Analytical studies have started to examine the optimal contract that aligns buyer-supplier incentives in joint NPD projects, but empirical evidence is scarce about the actual contracts offered by buying companies. Bridging the analytical and empirical literature, this paper compares optimal contracting derived from a parsimonious analytical model with actual behaviors observed in an experiment. In particular, we focus on how project uncertainty, buying company effort share, and buyer risk aversion influence three contractual decisions: total investment level, revenue share and fixed fee. Our results indicate significant differences between the optimal and actual behaviors. We identify various types of non-optimal contractual behaviors, which we explain from a risk aversion as well as a bounded rationality perspective. Overall, our findings contribute to the literature by showing that (1) the actual contractual behaviors could differ significantly from the optimal ones, (2) the actual contract design is sensitive to changes in project uncertainty and buying company effort share, and (3) the significant roles of risk aversion and bounded rationality in explaining the non-optimal contractual behaviors.