We use administrative data on the population of New Zealand construction firms from 2001-2012, along with linked data on their employees and working proprietors, to study the relationships among worker flows, entry, and firm productivity. We find that job churn is prevalent in construction, with around 60 percent of firm-worker pairs not existing previously or not existing subsequently. Firms with new employees are more productive than those with no change in workforce, in part because of knowledge flows from other construction firms. In our preferred specification, with firm fixed effects, a standard deviation increase in the productivity of new employees’ previous firms is associated with a 0.6 percent increase in productivity. Entrants are more productive than pre-existing firms. Firms that enter briefly and disappear exhibit high productivity for that brief period, and firms that enter and stay exhibit a persistent productivity advantage that averages about 6 percent, but which grows as experience accumulates. The entry and worker-knowledge-flow phenomena are distinct, in that the entry effect is not explained by employee composition, and non-entrant firms also benefit from worker knowledge flows.