OBJECTIVE: Glycemic control is a clinical outcome variable often used in diabetes intervention research. Misconceptions about A1C (glycosylated hemoglobin) and GHb (glycated hemoglobin) measures can lead to the use of different glycemic control measures within a study, threatening reliability and validity of outcomes results. While A1C is the standard for long-term glycemic control, the GHb measure may be more appropriate among patients with hemoglobin variants like the sickle cell factor and is sometimes used by clinicians who are aware that their patient population has prevalence of these variants. GHb measures are generally higher than A1Cs. The objective of this study was to examine the impact on outcomes results of using data from mixed (A1C or GHb) sites. METHODS: A convenience sample of 183 patients from four patient care sites participated in a cross-sectional survey measuring self-care behaviors, diabetes history, and demographics. Glycemic control data were gathered from patient charts and aggregated; one site used GHb. Aggregated glycemic control data were analyzed with other study variables. The GHb data were then converted to A1C equivalents using a regression formula provided by the National Glycohemoglobin Standardization Program (NGSP); analyses were conducted again and compared. RESULTS: Mean glycemic control for the sample with unconverted GHb readings was 8.40% (sd = 2.79); the mean for the A1C converted sample was 7.84% (sd=1.89). Significance of variable relationships varied between the two, and the magnitude of relationship among correlated variables was greater among the unconverted GHb sample. Mean glycemic control for both analyses revealed statistically significant associations (p < .05) with readiness for all diabetes self-care behaviors except vigorous exercise, but variable associations among the group with unconverted GHbs were stronger. CONCLUSIONS: If data within a study include different or non-standardized glycemic control measures, they should be converted to be sure results reflect a standard variable.