Rod produced from copper melted in a shaft furnace provides an excellent means of monitoring cathode quality if the feed is segregated according to source and correlated with output. While this approach is feasible in some fully integrated operations, specific cathode sampling procedures must be used in the case of direct cathode sale. By careful sampling and analysis of input to and output from a rod casting operation a more quantitative understanding of the requirements and limitations for sampling cathode lots was developed. The cathodes were sampled by drilling, by saw-cutting, and by cutting cathode strips. Statistical analysis of the data demonstrated the superiority of saw-cut samples over drill samples for predicting impurity concentrations in cathode lots.