The research was carried out in ten Seed Analysis Laboratories, located in the State of São Paulo, Brazil. It was used 10 commercial seed lots of guinea grass. The statistical analysis showed that when the tests were carried out in only one laboratory there were not significative difference between the results. However, in a lot, the difference between the results exceed the tolerance for the average betweeen the smallest and the greatest result. When the testa were carried out in different laboratories the resulta variation coefficients increased. The homogeneity of the purity analysis results did not improved the homogeneity of the germination tests results when these were carried out in different laboratories or even when they were carried out in only one.