Massive cyanobacteria blooms occur almost every summer in the Baltic Sea but the capability to quantitatively predict their extent and intensity is poorly developed. Here we analyse statistical relationships between multi-decadal satellite-derived time series of the frequency of cyanobacteria surface accumulations (FCA) in the central Baltic Sea Proper and a suite of environmental variables. Over the decadal scale (similar to 5-20 years) FCA was highly correlated (R-2 similar to 0.69) with a set of biogeochemical variables related to the amount of phosphorus and hypoxia in bottom layers. Water temperature in the surface layer was also positively correlated with FCA at the decadal scale. In contrast, the inter-annual variations in FCA had no correlation with the biogeochemical variables. Instead, significant correlations were found with the solar shortwave direct flux in July and the sea-surface temperature, also in July. It thus appears that it is not possible to predict inter-annual fluctuations in cyanobacteria blooms from water chemistry. Moreover, environmental variables could only explain about 45% of the inter-annual variability in FCA, probably because year-to-year variations in FCA are significantly influenced by biological interactions.