A model defining carbon isotope discrimination (delta13C) for crassulacean acid metabolism (CAM) plants was experimentally validated using Kalanchoe daigremontiana. Simultaneous measurements of gas exchange and instantaneous CO2 discrimination (for 13C and 18O) were made from late photoperiod (phase IV of CAM), throughout the dark period (phase I), and into the light (phase II). Measurements of CO2 response curves throughout the dark period revealed changing phosphoenolpyruvate carboxylase (PEPC) capacity. These systematic changes in PEPC capacity were tracked by net CO2 uptake, stomatal conductance, and online delta13C signal; all declined at the start of the dark period, then increased to a maximum 2 h before dawn. Measurements of delta13C were higher than predicted from the ratio of intercellular to external CO2 (p(i)/p(a)) and fractionation associated with CO2 hydration and PEPC carboxylations alone, such that the dark period mesophyll conductance, g(i), was 0.044 mol m(-2) s(-1) bar(-1). A higher estimate of g(i) (0.085 mol m(-2) s(-1) bar(-1)) was needed to account for the modeled and measured delta18O discrimination throughout the dark period. The differences in estimates of g(i) from the two isotope measurements, and an offset of -5.5 per thousand between the 18O content of source and transpired water, suggest spatial variations in either CO2 diffusion path length and/or carbonic anhydrase activity, either within individual cells or across a succulent leaf. Our measurements support the model predictions to show that internal CO2 diffusion limitations within CAM leaves increase delta13C discrimination during nighttime CO2 fixation while reducing delta13C during phase IV. When evaluating the phylogenetic distribution of CAM, carbon isotope composition will reflect these diffusive limitations as well as relative contributions from C3 and C4 biochemistry.