Context. Despite compliance with quality control standards, concerns remain as to the accuracy and reliability of point-of-care testing. Objective. To assess a practical method for quality improvement using the context in which point-of-care testing is done. Design. Quality measures for point-of-care testing, making use of natural duplication of results obtained by other testing methods, were used to monitor testing quality and evaluate quality improvement interventions. Setting. Five adult intensive care units (total of 88 beds) in a large academic medical center, using point-of-care testing for blood gases, electrolytes, and hematocrit levels. Participants. Nurses performing bedside testing and laboratory personnel assigned the responsibility for supervising their performance. Interventions. Quality of testing was monitored continuously, and, where problems were identified, training and support interventions implemented, and their effects evaluated. Main Outcome Measures. Improvement in correlation coefficients and regression parameters of point-of-care hematocrit and potassium testing results compared with contemporaneous results from the core laboratory. Results. The initial survey found point-of-care potassium levels were tightly correlated with core laboratory results (r = 0.958). Baseline correlation coefficients and regression parameters for point-of-care hematocrit levels compared with core laboratory values varied widely from unit to unit. The intensive care units with the highest variances of bedside vs core laboratory testing received targeted interventions. Follow-up yielded evidence of dramatic improvement; 1 unit experienced an increase in correlation from 0.50 to 0.95. Conclusions. The findings suggest that, when point-of-care testing is highly dependent on operator technique, targeted interventions can resolve problems and provide reliable results at the bedside.