Photofermentative hydrogen production from synthetic mixtures of lactose and lactate mimicking cheese whey was modeled and optimized using Design of Experiments and Response Surface Methodology. Five continuous parameters (light intensity, pH, lactose, lactate and glutamate concentrations) were studied as a function of buffer type (KPi or Borax) using two recombinant bacterial strains. For Rhodobacter capsulatus B10(lacZ), buffer type influenced the optimal parameter values but the optimal responses were similar in both buffers. In contrast, for R. capsulatus IR3(lacZ), responses were higher in Borax buffer than in KPi and were significantly higher than in strain B10(lacZ). Thus, the experimental optimized responses for specific volumetric H2 production, volumetric H2 production rate and substrate (lactose plus lactate) to H2 conversion rate in Borax buffer, were 12,150 ml L−1, 48.5 ml L−1 h−1 and 41.2%, respectively, for IR3(lacZ) compared to 6150 ml L−1, 33.5 ml L−1 h−1 and 32.5%, respectively, for B10(lacZ).