The purity P of laboratory chemicals is often declared in the form P > or = xy% (e.g., P > or = 97%). With a randomly chosen set of 40 compounds we found that their purity is generally closer to 100% than to the lower limit. The distribution of the purity data as found in the laboratory depends on the analytical technique used. Whereas purities determined by chromatography do not exceed 100% (because the sum of all observed peak areas is set to 100%), the purities obtained by titration can exceed 100% (because the functionality of the compound is measured). Therefore, the data for these two groups need to be dealt with in different ways. For purities based on titration we propose to use a rectangular distribution with a range from Pmin to 101%, an expected purity value which is the mean and a standard uncertainty of the purity u(P) of 29% of the range. Purities determined by chromatography can be described with a triangular distribution (ramp function). One leg of the triangle represents the range from Pmin to 100% and the right-angle is located at 100%. The expected value is the median and the uncertainty u(P) is 24% of the range. These proposals match the experimental data well.