For the quantification of particulate matter (PM) emitted into the atmosphere from industrial plants, reliable sampling methods are required. We have studied systematic errors in PM emission measurements performed according to the standard EN 13284-1, the legally required method in Europe for PM measurements of low concentrations in ducted gaseous streams. We analyzed a unique data set, covering 1614 PM samplings performed on a stack simulator, where the actual PM concentration was determined independently and traceably, allowing us to gain insight not only into the variability of the results obtained, but also into trueness of the measured PM concentrations. We found that the European Standard Reference Method (SRM) EN 13284-1 as applied by practitioners underestimates PM concentrations systematically by 13%. By varying the suction velocity into the sampling nozzles, as well as the nozzle geometry, we determined two main contributions for the underreporting of PM concentrations. First, the widespread practice by operators to perform PM sampling at 5% superisokinetic suction velocity leads to lower measured PM concentrations, which accounts for ca. one third of the underestimate. And second, the bulkiness of the nozzles generally used, which are not thin walled enough to allow representative sampling, with the use of bulky nozzles accounting for additional one third of the underreporting. • Systematic biases of the standard EN 13284-1 were studied at a stack simulator. • Dust sampling in stack emissions using this standard underreports particulate matter mass concentrations by 13%. • The generous limits on the deviation from isokinetic sampling are a major cause for underreporing. • The geometry of the sampling nozzles is another main cause for systematic underreporting.