Reconciliation between two copper ore mines transferred ore from one mine to another for processing in enrichment plants generated the need to regularly study the amount and composition of the ore on the conveyor connecting these two mines. To ensure the objectivity of the study, taking composite samples and their analysis was entrusted to a specialized outside laboratory. However, the managing staff of both mines still have doubts whether sampling results reflect correctly content of transported ore especially when the fed is highly variable. In order to investigate how the relatively low sampling rate affects the accuracy and precision of the measurement, the article investigates the hypothetical situation on the linking conveyor with the ore having extremely differentiated mineralization: 80% of almost barren rock (below 0.7% Cu) and 20% of the richly mineralized shale (around 10% Cu). Such ore occurs in some areas of the mine, from which it is fed onto a connecting conveyor. Through simulation techniques it was examined how the frequency of sampling can influence the distribution of the pooled sample results. It turned out that for 16 randomly selected samples in the following 15 minutes time intervals of a working shift, the spread of results around the simulated value is very large. A satisfactory accuracy level for the estimations of mean Cu content in the transported ore is achieved when the samples are collected at 30-second intervals. Only with sampling frequency close to on-line scanner parameters the probability of obtaining estimations with deviation exceeding 10% drops to the level of 2%. In the case of extremely differentiated ore doubts about confidence in the described measurements are fully confirmed, because with over 50% probability a single measurement could be deviated by 50% up and down from the true value.