The need for an integrated risk assessment at an ecologically relevant scale (e.g., at the population/community levels) has been acknowledged. Multispecies systems with increased ecological complexity, however, are difficult if not impossible to reproduce. The laboratory‐scale microcosm TriCosm (Pseudokirchneriella subcapitata, Ceriodaphnia dubia, Hydra viridissima) of intermediate complexity was developed for the reproducible assessment of chemical effects at the population/community levels. The system dynamics were repeatable in the short term, but interexperimental variation of algal dynamics in the long term triggered knock‐on effects on grazer and predator populations. We present 20 experiments to assess the effects of 12 factors (test medium, vessel type/condition, shaking speed, light intensity/regime, inoculation density, medium preparation components, metal concentration/composition, buffering salt type/concentration) on algal growth in the TriCosm enclosure. Growth rates varied between ≤ 0 and 1.40 (± 0.21) and generally were greatest with increased shaking speed, light exposure, medium buffer, or aeration time. Treatments conducted in dishes with aseptically prepared, lightly buffered, and/or hardly aerated medium resulted in low growth rates. We found that inter‐experimental variation of algal dynamics in the TriCosm was caused by a modification of medium preparation (omission of medium aeration) with the aim of reducing microbial contamination. Our findings highlight the facts that consistency in experimental procedures and in‐depth understanding of system components are indispensable to achieve repeatability. Environ Toxicol Chem 2019;00:1–13. © 2019 SETAC