A computational analysis of temporal patterns of acoustic intensity is applied to a series of recorded electroacoustic works in order to discern whether there are recurrent patterns in the intensity rise–fall structure. The works range from 1962 (Xenakis) to 2001 (Normandeau), and include composition and improvisation, large-scale (>25 minutes) and small-scale (<2 minutes) pieces, and acoustic, electronic and manipulated and synthesised sound. Contrary to Huron’s 1991 finding of long and gradual crescendi compared to short and abrupt decrescendi in classical music notation, it is found that in successive rise–fall events the rise phase (crescendo) is commonly shorter than the fall, and has a greater absolute rate of change of intensity than the fall. Correspondingly, crescendi occupy a smaller portion of the piece than decrescendi. When categorised into dynamic steps (such as p, mp, mf etc.), crescendi are usually less numerous than decrescendi. The results are considered from a cognitive point of view, in that this structuring may influence and thus result from listener attention and arousal patterns; and from a creator/performer point of view, in that embodied or virtualised effort may be key to the impact of an electroacoustic piece.