Cold pools are mesoscale features that are key for understanding the organization of convection, but are insufficiently captured in conventional observations. This study conducts a statistical characterization of cold-pool passages observed at a 280-m-high boundary layer mast in Hamburg (Germany) and discusses factors controlling their signal strength. During 14 summer seasons 489 cold-pool events are identified from rapid temperature drops below 22K associated with rainfall. The cold-pool activity exhibits distinct annual and diurnal cycles peaking in July and midafternoon, respectively. The median temperature perturbation is -3.3K at 2-m height and weakens above. Also the increase in hydrostatic air pressure and specific humidity is largest near the surface. Extrapolation of the vertically weakening pressure signal suggests a characteristic cold-pool depth of about 750 m. Disturbances in the horizontal and vertical wind speed components document a lifting-induced circulation of air masses prior to the approaching cold-pool front. According to a correlation analysis, the near-surface temperature perturbation is more strongly controlled by the pre-event saturation deficit (r = -0.71) than by the event-accumulated rainfall amount (r = -0.35). Simulating the observed temperature drops as idealized wet-bulb processes suggests that evaporative cooling alone explains 64% of the variability in cold-pool strength. This number increases to 92% for cases that are not affected by advection of midtropospheric low-Qe air masses under convective downdrafts.