This paper deals with the heterogeneous employment outcome at the plant level in Swedish manufacturing over the period 1972-96. <p> Non-negligible gross flows of jobs is found to be a pronounced feature in Swedish manufacturing, but as compared to results on U.S. data, the average pace of job reallocation has not been as high. However, masked behind low averages are periods of large-scale job reallocation and, in particular, we find that job reallocation exhibits a countercyclical movement. <p> Little of the observed heterogeneity in the plant-level employment outcome can be explained by easily observable characteristics of the plant. Instead most job reallocation takes place within narrowly defined sectors of the manufacturing sector. Furthermore, the role for idiosyncrasies in explaining the plant-level employment outcome becomes increasingly important in times of contraction. <p> We find no evidence supporting the hypothesis that large wage compression explains high job reallocation rates. Investigating the covariance structure of job reallocation, we instead find that, beside the net employment growth, the growth in productivity is the single most influential factor. <p> These findings, we like to believe, are consistent with theoretical models, which stress that the process of growth and technology adoption involves a great deal of experimentation. Accordingly, we find that these reallocative activities have been important in accounting for the long-run growth in productivity.