Because GC pairs are more stable than AT pairs, GC-rich sequences were proposed to be more adapted to high temperatures than AT-rich sequences. Previous studies consistently showed positive correlations between the growth temperature and the GC contents of structural RNA genes. However, for the whole genome sequences and the silent sites of the codons in protein-coding genes, the relationship between GC content and growth temperature is in a long-lasting debate. With a dataset much larger than previous studies (681 bacteria and 155 archaea), our phylogenetic comparative analyses showed positive correlations between optimal growth temperature and GC content exists, both in the structural RNA genes of bacteria and archaea and in bacterial whole genome sequences, chromosomal sequences, plasmid sequences, core genes, and accessory genes. However, in the 155 archaea, we did not observe a significant positive correlation of optimal growth temperature with whole-genome GC content or GC content at four-fold degenerate sites. We randomly drew 155 samples from the 681 bacteria for 1000 rounds. In most cases (> 95%), the positive correlations between optimal growth temperature and genomic GC content became statistically nonsignificant (P > 0.05). This result suggested that the small sample sizes might account for the lack of positive correlations between growth temperature and genomic GC content in the 155 archaea and the bacterial samples of previous studies.