Normal ranges for laboratory tests are customarily determined by results of test values from healthy persons. Patients’ laboratory test results recently have been analyzed by statistical and computer technics in an effort to obtain indirect estimates of the normal ranges. The present study was designed to examine the accuracy of normal ranges estimated indirectly by the following methods: normal probability graphs, truncated normal graphs, Pryce’s convention, average of normals unadjusted and adjusted procedures, and composite normal distributions. The estimated ranges were evaluated using the normal ranges determined concurrently on healthy adults as a standard of reference. The results revealed that the indirect estimates based on test results from adult hospital patients were quite inaccurate, because the estimated means were shifted significantly toward pathologic values and the estimated standard deviations were unacceptably wide.