The clinical usefulness of glycated hemoglobin (HbA1C) depends crucially on the accuracy and precision of its assay. When we compared an immunological bench-top analyzer (DCA 2000, Bayer Diagnostici, Milan) to the high-performance liquid chromatography (HPLC) reference method used in a routine hospital laboratory (Diamat and Fast Diamat, Bio-Rad Lab., Milan) by assaying multiple control sera, we found so many sources of systematic analytical errors in the routine use of HPLC as to compromise between-assay precision. DCA 2000 showed intra- and interassay coefficients of variation (CV) of 1.1% and 2.3% with the normal standard serum, 1.0% and 4.2% with the pathological one; Diamat yielded CVs of 1.3% and 7.0%, 1.3% and 5.7%, respectively. Although the measurement of 161 blood samples showed that Diamat usually overestimated HbA1C (paired t-test, P<0.001), a great variability of Diamat performance became evident when the relationship Diamat vs DCA was evaluated day by day over 17 days of observation (analysis of variance, ANOVA, P<0.001). Intra- and interassay CVs of Fast Diamat initially (new instrument still on approval) were 0.6% and 2.5% (normal standard serum), 0.3% and 1.9% (high standard serum), yet after 6 months of routine laboratory use, they became 3.1% and 3.2%, 1% and 12.3%, respectively. Main sources of error were: inaccurate autodilution, unsuitable parameter settings, disregard of the maintenance schedule. We conclude that the tendency to overlook major critical aspects in the routine use of HPLC is detrimental to the quality of HbA1C determination and implies the loss of HbA1C value in clinical practice. Both carefully supervising laboratory quality and checking the likelihood of the analytical result with the clinical setting appear even more important.