Purpose: To evaluate whether retention loss of Class V restorations can be simulated in the laboratory and to compare these results with those from clinical trials. Materials and Methods: Nonretentive v-shaped Class V cavities were prepared on the lingual and buccal side of extracted premolars, half in dentin and half in enamel. Different adhesive systems (AS) were used with the same composite resin (Tetric EvoCeram) and 12 restorations per group: 1-step self-etching AS (AdheSE One, Adper Prompt L-Pop, Futurabond N, Hybrid Bond, iBond, Xeno III, Xeno IV), 2-step self-etching AS (AdheSE), 2-step etch-and-rinse AS (Excite, Prime & Bond NT), 3-step etch-and-rinse AS (Syntac) as well as a conventional glass ionomer (Ketac Fil) with and without conditioner. The comparison groups were the composite without adhesive and a 2-step etch-and-rinse AS (Prime & Bond NT) without etching of enamel and dentin. The restored teeth were submitted to an aging process involving 18 months of water storage, three intermittent phases of thermocycling (TC 10,000 times), and two phases of thermomechanical loading (1 x 640,000 chewing cycles after 12 months, 1 x 1,200,000 chewing cycles after 18 months; 100 N sine-wave force profile, pressing with steel ball without lifting). Retention loss of the restorations was evaluated after every 1000 thermocycles and every 120,000 cycles of thermomechanical loading. The databases MEDLINE and IADR abstracts were used to search for clinical studies on retention loss involving the adhesive systems that were included in the present study. Results: Retention loss was only observed in the following groups: composite without adhesive (100% after first 1000 TC), glass ionomer without conditioner (8% after 6 months; 33% after 12 months, 100% after 18 months), adhesive without etching (17% after 6 months, 42% after 12 months). The laboratory results, however, matched with the clinical results only for three adhesive systems (Futurabond NR, Hybrid Bond, Xeno IV, 0% retention loss, 5 studies, observation period between 1.5 and 2 years). Conclusions: If the materials were applied according to the manufacturer's instructions, no retention loss was observed in the laboratory model. The laboratory model did not reflect the clinical findings. Schlagwörter: Class V, retention loss, laboratory, water storage