A major route for hydrogen production by purple photosynthetic bacteria is biological nitrogen fixation. Nitrogenases reduce atmospheric nitrogen to ammonia with the concomitant obligate production of molecular hydrogen. However, hydrogen production in the context of nitrogen fixation is a rather inefficient process because about 75% of the reductant consumed by the nitrogenase is used to generate ammonia. In this study we describe a selection strategy to isolate strains of purple photosynthetic bacteria in which hydrogen production is necessary for growth and independent of nitrogen fixation. We obtained four mutant strains of the photosynthetic bacterium Rhodopseudomonas palustris that produce hydrogen constitutively, even in the presence of ammonium, a condition where wild-type cells do not accumulate detectable amounts of hydrogen. Some of these strains produced up to five times more hydrogen than did wild-type cells growing under nitrogen-fixing conditions. Transcriptome analyses of the hydrogen-producing mutant strains revealed that in addition to the nitrogenase genes, 18 other genes are potentially required to produce hydrogen. The mutations that caused constitutive hydrogen production mapped to four different sites in the NifA transcriptional regulator in the four different strains. The strategy presented here can be applied to the large number of diverse species of anoxygenic photosynthetic bacteria that are known to exist in nature to identify strains for which there are fitness incentives to produce hydrogen.