Tree stems contract and expand as stem water is depleted and replaced. Band-dendrometer studies suggest that such daily changes are small (<0.2 mm diameter), and they are ignored in most growth measurements. However, several studies using other approaches note larger changes (even >1 cm diameter), suggesting that significant biases are possible. An exploratory study examined the pattern and magnitude of daily stem changes and whether commercial band-dendrometers were able to reveal them. A method involving multiple precision measurements on eight trees in a Bornean hill dipterocarp forest revealed daily shrinkage and expansion of girth of around 1 mm. Fluctuations were greater in bright weather. Band-dendrometers detected these changes but revealed less than a tenth of their magnitude. An analytical model for dendrometer error is presented that predicts how measurement biases can be reduced. Tropical trees can fluctuate appreciably in stem diameter over the day. These reversible changes are of sufficient magnitude to merit concern in growth studies. Influential biases seem especially likely when measurement intervals are short and involve systematic differences in timing and weather. Further study is required to gauge the more general influence of these measurement problems.