Many articles dealing with theoretical and observational aspects of atmospheric blocking have appeared in the literature. So far little attention has been paid to the question of long term variations in blocking frequency. Usually 500-hPa geopotential height data are used in observational studies of atmospheric blocking. We used a 94-year time series of sea-level pressure data in order to extend the analysis back to the first half of the century.Long term variations of frequency of blocked flow as well as individual blocking events have been documented. There are two areas of study, the North Atlantic and Western Europe (400W to 300E) and the Pacific (1600E to 1300W). Low-pass filtered data suggest that there was increased blocking activity during the winter and spring seasons in the Atlantic during 1930-1935 and 1950-1960, whereas there was reduced blocking activity during 1910-1915, in particular during the winters. However, a statistical test did not verify that these peaks are real. Data for the Pacific do not show any outstanding periods with blocked flow.Another result is that there are less blocking episodes in the Pacific than in the Atlantic. Data for the whole time period show that the number of episodes in the Pacific is 60% of the number of episodes in the Atlantic during winter and spring. During summer and autumn there are less blocking episodes in both regions than during the other two seasons. Furthermore, the number of episodes in the Pacific is only 40% of the number of episodes in the Atlantic. We also found that there is a factor of almost 2 between the number of blocking episodes during those decades with the highest number of blocking episodes and those with the lowest. This is true for both the Atlantic and the Pacific.