This paper investigates the quantitative effects of employee turnover on firms' productivity.The Australian Business Longitudinal Survey 1995-98, a unique survey providing firm level data on both production and employee turnover, is used as the data source. Theoretical studies have advocated that firm specific human capital and job matching to be the two major, but competing, mechanisms through which turnover affects productivity. Our results indicate that the effect of job matching dominates when turnover is"low," while the effect of firm specific human capital dominates when turnover is"high." We identify that the optimal turnover rate - the rate that maximises productivity, controlling for other factors - is about 0.3, well in excess of the sample mean. The finding suggests that further increasing the flexibility of employment arrangement for small and medium Australian enterprises could yield substantial productivity gains.