In the Himmerfjärden Bay a large excess of nitrogen over phosphorus in the discharge from a large sewage treatment plant (STP) has suppressed growth of diazotrophic cyanobacteria in its inner parts. Implementation of nitrogen removal in the STP in 1997 drastically reduced nitrogen load and triggered growth of diazotrophs, mainly Aphanizomenon sp. This study is part of a long-term series of experiments with the overall aim to test how algal biomass and production in a receiving area can be reduced, without stimulating nitrogen fixation and biomass growth by diazotrophic cyanobacteria. Nitrogen removal was discontinued in the STP during two years (2007-8) and resumed in 2009, and the discharge shifted from 25 to 10 m depth, above the seasonal pycnocline. Cellular 15N showed that N2 was the most important N source for diazotrophic cyanobacteria, and that uptake of combined nitrogen was insignificant. As biomass was declining and at the end of the productive season, we could detect a small, but significant, increase in cellular δ15N at the inner bay stations (H3 and H4). However, this coincided with an increased proportion of Anabaena spp. of the total diazotrophic biomass. This may indicate that Anabaena spp. has a higher uptake of combined nitrogen compared with Aphanizomenon sp. or that declining populations of Aphanizomenon sp. take up combined nitrogen. We also found no evidence of uptake of combined nitrogen during the winter months when nitrogen supply is ample and Aphanizomenon sp. is devoid of heterocysts. During the first half of summer (week 21-27) heterocyst frequencies were higher at the outer stations B1 and H2, compared to the inner bay stations (H4 and H5). The lower frequencies at the inner bay stations are likely due to the reduced growth rate suffered by the Aphanizomenon sp. due to stronger competition for phosphorus by non-diazotrophs at these stations and hence lower need for heterocysts. Towards the end of summer conditions even out along the bay, as the surplus phosphorus from the spring bloom is used up at the outer stations and no heterocyst gradient is present. Heterocyst frequency varied significantly over the summer, with minimum values in the beginning of July, coinciding with the highest water temperatures, and higher frequencies in early and late summer. We suggest this is primarily due to a more efficiently functioning nitrogenase enzyme at high temperatures with a reduced need for “expensive” heterocysts. Spring heterocyst differentiation occurred around 4-6 weeks after depletion of dissolved inorganic nitrogen (DIN) and only when water temperature was 5-9 oC and a pycnocline established. It seems that temperature and light in concert will initiate growth, leading to an internal nitrogen deficiency which starts heterocyst differentiation.