Objective: The aim of the work was the comparison of precision under conditions of repeatability two reconstituted control human materials from 4 different analytical systems (Kone – Konelab 60, Arkray – Spotchem EZ, Abaxis – Piccolo and Ortho-Clinical Diagnostics – Vitros 950). At these analyzers was estimated 12 basic biochemical analytes (albumin, ALP, ALT, AMS, AST, bilirubin, urea, total calcium, cholesterol, creatinine, glucose, total protein). Judgment of comparability of results coming through different analytical systems (two small POCT analysers: Piccolo – “wet chemistry“ and Spotchem – “dry chemistry “ and two big biochemical analyzers with big capacity: Vitros 950 – “dry chemistry analyser“ and Konelab 60 – classical “wet chemistry“ analyser. Material and Methods: All estimations were conducted according to producer recommendations and using the original producer recommended reagents. For evaluation of repeatability were analysed commercial control materials (lyophilized human sera after reconstitution). Results: Repeatability expressed as coefficient of variation achieved typical value about 2.5% with maximum 5% with exceptions as follows: ALP/Spotchem (14.9% and 8.1%) – trend in data, ALT/Piccolo (18.0%) – outlier, urea/Piccolo (7.4%) – generally great variance of values, cholesterol/Piccolo (7.6% and 6.0%) – abnormality of data and outlier. Reproducibility expressed as coefficient of variation was typically to 7%, higher values (7–11%) were found as follows: ALT/Piccolo, bilirubine/Kone, urea/Kone, Ca/Kone, kreatinin/Kone. The bias value was posible count only for calcium, cholesterol, glucose and creatinine. Statistical significant bias (significance level 0,05) indicates Ca/Spotchem (-12% and -15%), cholesterol/Vitros (+5%), glucose/Kone (+5%), glucose/Spotchem (-12%), creatinin/Piccolo (+8%), kreatinin/ Vitros (+15% and +14%). Results comparison of patiens plasma and serum in all analysers is demonstrated at figures. Conclusion: Repeatability was good, coefficient of variation was max. 5%. Reasons of exceptions were caused by outliers or abnormal distribution (coeff. of variation 18% – ALT/Piccolo, 7.5% and 13.6% – cholesterol/Piccolo, 5.3% – AMS/ Spotchem) and by trends of data (coeff. of variation 14.9% and 8.1% – ALP/Spotchem). Only in one case (e.g. urea/Piccolo) was generally very large values distribution – coeff. of variation 7.4%. Reproducibility as coefficient of variation was found up to 7%, exceptionally up to 11%. Because of the reason of the small cohort of data (n = from 3 to 5) it is neccessary to evaluate reproducibility very carefully. Each of four analytes (total calcium, cholesterol, glucose, creatinine) for which was possible to evaluate bias appeared minimally on one analyte, at least in one of four compared analyzers, statistically significant bias in range from -15% up to +16%. In random cross comparison of achieved results were remarkably incomparable results in following cases: ALT/Piccolo/plasma – proportional negative bias (influence of calibration?); AMS at all 5 cases – generally poor comparability (methods heterogeneity?); bilirubine at all 7 cases – in range from 5 up to 15 μmol/l was generally found bad comparison (low sensitivity?); glucose/Spotchem/plasma and serum – not big but significant negative additive bias (blank?); total protein/Kone and Vitros in serum indicates no big, but significant additive bias against values measured on Kone and Vitros in plasma (influence of fibrinogen?).

        Key words: repeatability, Piccolo, Spotchem, Vitros 950, Konelab 60, dry chemistry, wet chemistry