The Temperament and Character Inventory (TCI) was developed to measure the following temperament dimensions: novelty seeking (NS), harm avoidance (HA), reward dependence (RD) and persistence (P). These four dimensions of temperament were originally proposed to be independent of one another. In this study the inter-relationships between the dimensions were studied with meta-analytic techniques. We also studied the effects of sociodemographic factors (location of the study, mean age and gender distribution) on correlations between temperament dimensions. We searched studies on healthy (non-clinical) populations that used the TCI (version 9), and that had a required sample size of at least 100. The search resulted in 16 articles. The resulted pooled correlation coefficient was medium level between NS and HA (-0.27). Correlations were small for HA-P (-0.20), NS-P (-0.14), NS-RD (0.10), RD-P (0.05) and HA-RD (0.04). In meta-regression, the correlation NS-P was significantly affected by the location of the study (Asian/other) and by the gender distribution of the sample. In the HA-P correlation, the mean age of the sample affected the correlation. In conclusion, we found a medium level negative correlation between NS and HA; other correlations between the dimensions were small. These findings mainly support Cloninger's theory of independent dimensions.