Because of the work of the pioneers of sampling theory, we know the principles, test work and equations to use to design sampling systems to meet targets for precision and accuracy. However, we do not see such proper' systems deployed everywhere a sample is obtained around mineral processing plants. In many cases, we do not have solutions that satisfy all of the theoretical, economic and practical considerations. There are some well-known problems, such as high-volume streams with wide size distributions combined with a sparse valuable component that is heterogeneously distributed (such as gold ore SAG mill feeds). However, other difficult sampling issues are common, such as streams which vary widely over a day in flow rate and slurry density (many flotation concentrates), and loaded CIP carbon on transfer to elution. Inventory measurement issues abound, especially for leaching operations where daily inventory change is significant for metallurgical accounting. The result is that operators realise that although they issue reports daily, they do not know what is actually going on in their plant on a daily basis with sufficient confidence to quickly diagnose problems or confidently predict the benefits of potential enhancements. The industry needs practical and economic solutions that can be awarded the theoretically correct sampler' tick in order to better understand and optimise their processes.