The reliability of measurements of mainstream smoke analytes other than tar, nicotine and CO is not known but is important in the current regulatory environment internationally. An appreciation of between-laboratory variability is essential for companies contracting analytical work to outside suppliers. Six laboratories obtained smoke data from three cigarette brands for as many of the 44 'Hoffmann' analytes as they could currently measure. The brands (tar yields 12 mg, 8 mg, and 5 mg), were smoked under the ISO smoking regime to obtain yield values based on 5 replicates, each laboratory using their chosen number of cigarettes per replicate. In addition, laboratories used their preferred and internally validated methodology; i.e., smoking machine type, trapping system, sample work-up, and detection system. Around 2900 data points were obtained. This study was based on measurements at a single time point and did not include any components of longer-term variability that would be expected to increase further the measurement variability. No analytes had lower within-laboratory measurement variability than tar, nicotine and CO; 73% of the other analytes had statistically higher levels. All laboratories ranked the products in the same order for all analytes (except some metals) but there was as much as 10-fold difference in measured values between laboratories. The mean variation between highest and lowest yield measurements was 60% when 3 values in excess of 8-fold were excluded. Given the lack of standardised methods, it is not currently possible to make meaningful comparisons between such data from several sources with this degree of inter-laboratory variability. Indeed, calculation of yields from benchmarking studies may prove more reliable.