Cassava brown streak disease (CBSD) has occurred in the Indian Ocean coastal lowlands and some areas of Malawi in East Africa for decades, and makes the storage roots of cassava unsuitable for consumption. CBSD is associated with Cassava brown streak virus (CBSV) and the recently described Ugandan cassava brown streak virus (UCBSV) [picorna-like (+)ssRNA viruses; genus Ipomovirus; family Potyviridae]. This study reports the first comprehensive analysis on how evolution is shaping the populations of CBSV and UCBSV. The complete genomes of CBSV and UCBSV (four and eight isolates, respectively) were 69.0-70.3 and 73.6-74.4% identical at the nucleotide and polyprotein amino acid sequence levels, respectively. They contained predictable sites of homologous recombination, mostly in the 3'-proximal part (NIb-HAM1h-CP-3'-UTR) of the genome, but no evidence of recombination between the two viruses was found. The CP-encoding sequences of 22 and 45 isolates of CBSV and UCBSV analysed, respectively, were mainly under purifying selection; however, several sites in the central part of CBSV CP were subjected to positive selection. HAM1h (putative nucleoside triphosphate pyrophosphatase) was the least similar protein between CBSV and UCBSV (aa identity approx. 55%). Both termini of HAM1h contained sites under positive selection in UCBSV. The data imply an on-going but somewhat different evolution of CBSV and UCBSV, which is congruent with the recent widespread outbreak of UCBSV in cassava crops in the highland areas (>1000 m above sea level) of East Africa where CBSD has not caused significant problems in the past.