Results are presented from the first investigation into a summer-season of mesoscale vortex activity in a large sector of the Antarctic coastal region. The study is based on an analysis of 3 months’ meteorological satellite imagery collected at the British Research Station ‘Rothera’ on the Antarctic Peninsula. The study revealed the high frequency with which such systems occur, with 162 individual vortices being found during the period December 1983 to February 1984 inclusive. The preferred area for their development was in the latitude band 60–70°S over the marginal ice zone and ice-free region of the eastern Bellingshausen Sea. A classification scheme for the vortices was developed based on the relationship with the broad-scale synoptic flow, the sea ice, and the geographical location. The most common type of vortex found was the ‘classic’ polar low, which formed in the southerly flow to the west of synoptic-scale disturbances. These vortices were very similar to the baroclinic type of polar lows observed south of Iceland during the Northern Hemisphere winter. A third of the vortices in total were found to be mesoscale features associated with synoptic-scale troughs or the centres of major depressions. Vortices with comma-shaped cloud signatures occurred about twice as frequently as those with spiraliform cloud. The vast majority of vortices had a diameter of less than 500 km, with very few systems being observed in the range 500–1000 km. Mean anomalies of 500 hPa geopotential height and surface pressure for the occasions when vortices were identified were —5.3 dm and — 0.5 hPa, respectively, indicating the association of these systems with upper air troughs and cold pools. Only 23 of the vortices found were correctly represented on the Meteorological Office analyses and of these 15 were small synoptic disturbances. The ‘polar low’ class of vortex was very poorly represented in the analyses, indicating that the available satellite sounder data could not resolve the systems and that the processes resulting in their formation were not handled well by forecast/data assimilation schemes. Comparison of the mean surface and 500 hPa height fields for this 3-month period with the long-term average data show that there were negative anomalies at both levels over the Bellingshausen Sea. The number of vortices over the Bellingshausen Sea in this summer period may have been greater than would be expected in an average year, but activity over the Weddell Sea was probably close to average.