Cyclical rise and fall of monthly mean CO2 concentration occurred rather regularly at Mauna Loa, Hawaii, USA, during 1958–2008, with peaks occurring in April or May followed by troughs in September or October. However, the frequency of troughs in September increased from 32% in 1958–1974 to 82% in 1992–2008 and the depletion season correspondingly decreased from 4.8 months on average to 4.2 months. Does this seemingly small change in the seasonal rhythm point to changes on an inter-decadal scale in the CO2 accumulation rates?