Capacity of a freeway segment should only be measured when it is an active bottleneck. The properties of flows at active freeway bottlenecks bear on both the definition of capacity and the procedure of capacity analysis. Past studies have examined the flow features at bottlenecks on several freeways in Toronto, Ontario, and San Diego, California. This study examined twentyseven active bottlenecks in the Twin Cities metro area, Minnesota, for a seven-week period. The analysis focuses on the properties of pre-queue transition flows (PQF) and queue discharge flows (QDF) averaged across various time intervals (thirty-second, daily average, and long-run average). It is found that the percentage flows drop after upstream queues form at all studied bottlenecks range from 2 percent to 11 percent. The 30-second QDFs display high variation and should not be assumed to be constant. The daily average QDFs at each studied bottleneck follows a normal distribution based on two normality tests and visual inspection of the normal probability plot. Results also suggest that the long-run average QDFs (mean 2016 pc/ln/hr) and PQFs (mean 2124 pc/ln/hr) are both normally distributed. The implication of these empirical findings on capacity estimation is also discussed.