A B S T R A C TSome recent publications promote one-run, open-system pyrolysis experiments using a single heating rate (ramp) and fixed frequency factor to determine the petroleum generation kinetics of source-rock samples because, compared to multiple-ramp experiments, the method is faster, less expensive, and presumably yields similar results. Some one-ramp pyrolysis experiments yield kinetic results similar to those from multiple-ramp experiments. However, our data for 52 worldwide source rocks containing types I, II, IIS, II/III, and III kerogen illustrate that one-ramp kinetics introduce the potential for significant error that can be avoided by using high-quality kinetic measurements and multiple-ramp experiments in which the frequency factor is optimized by the kinetic software rather than fixed at some universal value. The data show that kinetic modeling based on a discrete activation energy distribution and three different pyrolysis temperature ramps closely approximates that determined from additional runs, provided the three ramps span an appropriate range of heating rates. For some source rocks containing well-preserved kerogen and having narrow activation energy distributions, both single-and multiple-ramp discrete models are insufficient, and nucleation-growth models are necessary. Instrument design, thermocouple size or orientation, and sample weight likely influence the acceptable upper limit of pyrolysis heating rate. Caution is needed for ramps of 30-50°C/min, which can cause temperature errors due to impaired heat transfer between the oven, sample, and thermocouple. Compound volatility may inhibit pyrolyzate yield at the lowest heating rates, depending on the effectiveness of the gas sweep. We recommend at least three pyrolysis ramps that span at least a 20-fold variation of comparatively lower rates, such as 1, 5, and 25°C/min. The product of heating rate and