The poor record of basic-to-clinical translation in recent decades has led to speculation that preclinical research is "irreproducible", and this irreproducibility in turn has largely been attributed to deficiencies in reporting and statistical practices. There are, however, a number of other reasonable explanations of both poor translation and difficulties in one laboratory replicating the results of another. This article examines these explanations as they pertain to preclinical pain research. I submit that many instances of apparent irreproducibility are actually attributable to interactions between the phenomena and interventions under study and "latent" environmental factors affecting the rodent subjects. These environmental variables-often causing stress, and related to both animal husbandry and the specific testing context-differ greatly between labs, and continue to be identified, suggesting that our knowledge of their existence is far from complete. In pain research in particular, laboratory stressors can produce great variability of unpredictable direction, as stress is known to produce increases (stress-induced hyperalgesia) or decreases (stress-induced analgesia) in pain depending on its parameters. Much greater attention needs to be paid to the study of the laboratory environment if replication and translation are to be improved.