This paper presents a bootstrap approach to calculate confidence intervals for firm-specific Malmquist productivity indices obtained from data envelopment analysis (DEA) models. The bootstrap is easily implemented and allows identification of production units that have significant productivity changes. An application using data from Swedish eye-care departments is included. We find that 40% of departments have significant progress in productivity whereas only 10% of the departments have a significant regress in productivity. This differs from the original results where about half the sample have estimates of progress and the other half have estimates of regress in productivity.