Oil-prone marine petroleum source rocks contain type I or type 11 kerogen having Rock-Eval pyrolysis hydrogen indices greater than 600 or 300-600 mg hydrocarbon/g total organic carbon (HI, mg HC/g TOC), respectively. Samples from 29 marine source rocks worldwide that contain mainly type II kerogen (HI = 230-786 mg HC/g TOC) were subjected to open-system programmed pyrolysis to determine the activation energy distributions for petroleum generation. Assuming a burial heating rate of 1 degrees C/m.y. for each measured activation energy distribution, the calculated average temperature for 50% fractional conversion of the kerogen in the samples to petroleum is approximately 136 +/- 7 degrees C, but the range spans about 30 degrees C (similar to 121-151 degrees C). Fifty-two outcrop samples of thermally immature Jurassic Oxford Clay Formation were collected from five locations in the United Kingdom to determine the variations of kinetic response for one source rock unit. The samples contain mainly type I or type II kerogens (HI = 230-774 mg HC/g TOC). At a heating rate of 1 degrees C/m.y., the calculated temperatures for 50% fractional conversion of the Oxford Clay kerogens to petroleum differ by as much as 23 degrees C (127-150 degrees C). The data indicate that kerogen type, as defined by hydrogen is not systematically linked to kinetic response, and that index, default kinetics for the thermal decomposition of type I or type II kerogen can introduce unacceptable errors into numerical simulations. Furthermore, custom kinetics based on one or a few samples may be inadequate to account for variations in organofacies within a source rock. We propose three methods to evaluate the uncertainty contributed by kerogen kinetics to numerical simulations: (1) use the average kinetic distribution for multiple samples of source rock and the standard deviation for each activation energy in that distribution; (2) use source rock kinetics determined at several locations to describe different parts of the study area; and (3) use a weighted-average method that combines kinetics for samples from different locations in the source rock unit by giving the activation energy distribution for each sample a weight proportional to its Rock-Eval pyrolysis S-2 yield (hydrocarbons generated by pyrolytic degradation of organic matter).