The Belgian data (2003–2010) for the European Antimicrobial Resistance Surveillance Network (EARS-Net) showed a significant decreasing trend in the proportion of penicillin non-susceptible Streptococcus pneumoniae (9·4% to <1%) from blood and CSF isolates. We found that 75% of this decrease was explained by a change in Clinical and Laboratory Standards Institute (CLSI) breakpoints as the trend disappeared if only the new breakpoints were applied. Applying only European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints also resulted in a relatively stable proportion of penicillin non-susceptibility (average 5%), but this proportion was 7–13 times higher than with the new CLSI breakpoints. When the new CLSI breakpoints alone are used, fewer than 1% of bacteraemia isolates were penicillin non-susceptible during the entire period, but the proportion of non-susceptible meningitis isolates rose from 6·3% in 2003 to 15·9% between 2003 and 2010. Changing breakpoints should lead to retrospective analysis of historical data to minimize wrongly interpreting resistance trends.