Bursting bubbles are thought to be the dominant cause of cell death in sparged animal or insect cell cultures. Cells that die during the bubble burst can come from three sources: cells suspended near the bubble; cells trapped in the bubble lamella; and cells that attached to the rising bubble. This article examines cell attachment to rising bubbles using a model in which cell attachment depends on cell radius, bubble radius, and cell–bubble attachment time. For bubble columns over 1 m in height and without protective additives, the model predicts significant attachment for 0.5- to 3-mm radius bubbles, but no significant attachment in the presence of protective additives. For bubble columns over 10 cm in height, and without protective additives, the model predicts significant attachment for 50- to 100-μm radius bubbles, but not all protective additives prevent attachment for these bubbles. The model is consistent with three sets of published data and with our experimental results. Using hybridoma cells, serum-free medium with antifoam, and 1.60 ± 0.05 mm (standard error) radius bubbles, we measured death rates consistent with cell attachment to rising bubbles, as predicted by the model. With 1.40 ± 0.05 mm (SE) radius bubbles and either 0.1% w/v Pluronic-F68 or 0.1% w/v methylcellulose added to the medium, we measured death rates consistent with no significant cell attachment to rising bubbles, as predicted by the model. © 1999 John Wiley & Sons, Inc. Biotechnol Bioeng 62: 468–478, 1999.