The underlying reasons and treatment of sub-optimum performance (SOP) were assessed in a dairy herd. The need for intensive investigations, problems encountered with the use of published reference values, and the probability of delayed herd response to treatment are addressed. The study was done in a farm holding 55 milking and dry cows, with a history of sub-optimum milk production and reproduction. A series of blood and serum experiments were done on all cows and the results were compared with the midpoints of the corresponding reference values, using one sample t test. RBC and WBC were normal. Hemoglobin and PCV were within the reference intervals, close to the lower limit. Glucose was almost at the lowest limit of the reference interval, while triglycerides and cholesterol, respectively, were lower and higher than the limits. Hypoalbuminemia, hyper alpha- and gamma-globulinemia as well as normal levels of total serum proteins and beta-globulins were the other findings. According to the initial results, farm history and environmental assays, nutritional insufficiencies and sub-clinical infections were suspected as the underlying reasons of sub-optimum performance. The treatment procedure was designed based on the treatment of nutritional insufficiencies with the expectation of the self-cure of sub-clinical infections. The same experiments were done on the same cows on days 90 and 210 of the study. The changes were studied by the proc-mixed procedure and Tukey's multiple range test of SAS statistical software. RBC decreased (P<0.05) but WBC did not change (P>0.05) during the study. Hemoglobin and PCV increased (P<0.05). Serum glucose, triglycerides, cholesterol and albumin increased (P<0.05). Hyper alpha- and gamma-globulinemia were corrected and total protein and beta globulins decreased (P<0.05). These changes did not affect the daily milk average of the herd during the 7 months of the study as it remained almost constant. For investigating SOP, in the absence of local reference values, relying on published reference values may not help diagnosis. Treatment of SOP may not be reflected on herd production level in short terms and the results should be interpreted by care using a sort of herd records.