Reliable estimates of in situ phytoplankton growth rates are central to understanding the dynamics of aquatic ecosystems. A common approach for estimating in situ growth rates is to incubate natural phytoplankton assemblages in clear bottles at fixed depths or irradiance levels and measure the change in chlorophyll a (Chl) over the incubation period (typically 24 h). Using a modelling approach, we investigate the accuracy of these Chl-based methods focussing on 2 aspects: (1) in a freely mixing surface layer, the cells are typically not in balanced growth, and with photoacclimation, changes in Chl may yield different growth rates than changes in carbon; and (2) the in vitro methods neglect any vertical movement due to turbulence and its effect on the cells' light history. The growth rates thus strongly depend on the incubation depth and are not necessarily representative of the depth-integrated in situ growth rate in the freely mixing surface layer. We employ an individual based turbulence and photosynthesis model, which also accounts for photoacclimation and photoinhibition, to show that the in vitro Chl-based growth rate can differ both from its carbon-based in vitro equivalent and from the in situ value by up to 100%, depending on turbulence intensity, optical depth of the mixing layer, and incubation depth within the layer. We make recommendations for choosing the best depth for single-depth incubations. Furthermore we demonstrate that, if incubation bottles are being oscillated up and down through the water column, these systematic errors can be significantly reduced. In the present study, we focus on Chl-based methods only, while productivity measurements using carbon-based techniques (e.g. 14 C) are discussed in Ross et al. (2011; Mar Ecol Prog Ser 435:33-45).