BACKGROUND: Two-phase partitioning bioreactors (TPPBs) are based on the addition of a non-aqueous phase (NAP) to a biological process in order to overcome a limited delivery of gaseous substrates to the microorganisms in the case of compounds with low affinity for water. However, the high power input (P-g/V) required to disperse the NAP is often the major limitation for TPPB applications at full scale. Therefore, the accurate determination of the overall mass transfer coefficient (K(L)a) at low P-g/V values is a critical issue as these operational conditions are more attractive from a scale-up point of view. RESULTS: NAP addition altered the typical shape of the dissolved oxygen curves used for KLa determination at the lowest P-g/V values tested (70-80 W m(-3)). Below a threshold P-g/V value of 600 W m(-3), the presence of the NAP increased the error in K(L)a measurements up to 115% relative to controls deprived of NAP. CONCLUSIONS: The error in K(L)a measurements at low P-g/V values might be related to failures in the fundamental assumption regarding liquid phase homogeneity in the mass transfer model used. (C) 2010 Society of Chemical Industry