This study examines the extent to which parsimonious and general cross-sectional valuation models, restricted to include only publicly available historical accounting information, explain share prices in the cross section, identify periods when market mispricing may be more pervasive, and also identify which shares within those cross sections are more likely to be mispriced. Our model simply includes historical book value, earnings, dividends, and growth, but it explains on average over 60 percent of the cross-sectional variation in share prices in annual estimations across 1975-2011. We also examine the extent to which the residuals indicate mispricing. The quintile of stocks picked by our model as most likely underpriced outperform the quintile of stocks picked as most likely overpriced by an average of 9.9 percent over the following 12 months, after controlling for size. We also predict and find that value residuals are better predictors of future abnormal returns: (i) among firms that are not covered by analysts; (ii) among firms that face fewer accounting measurement challenges; and (iii) when we estimate value model parameters by industry/year. We also predict and find our approach works better in periods when the mapping of fundamentals into prices is weaker. This study contributes a novel and straightforward approach to map accounting fundamentals into share prices in order to identify mispricing in time-series and in the cross section. Evaluation et erreurs d' evaluation des param etres comptables fondamentaux dans la s erie chronologique et l' echantillon * Accepted by Partha S. Mohanram. We appreciate helpful comments and suggestions from the editor and two anonymous reviewers, as well as Tom Canace and Marilyn Johnson, and workshop participants at the estimations annuelles de la p eriode 1975-2011. Les auteurs se demandent egalement dans quelle mesure les valeurs r esiduelles r ev elent des erreurs d' evaluation. Les rendements du quintile des titres choisis, selon leur mod ele, comme etant les plus susceptibles d'être sous-e valu es exc edent de 9,9 pour cent, en moyenne, les rendements du quintile des titres choisis comme etant les plus susceptibles d'être sur evalu es, au cours des 12 mois suivant l'estimation du mod ele, une fois contrôl ee la taille. Les auteurs pr evoient, et constatent egalement, que les valeurs r esiduelles sont de meilleurs indicateurs des rendements futurs anormaux i) parmi les soci et es qui ne retiennent pas l'attention des analystes, ii) parmi celles pour qui l' evaluation comptable pr esente moins de difficult es et iii) lorsque les auteurs estiment les param etres du mod ele de la valeur selon le secteur et l'ann ee. Conform ement a leur hypoth ese, les auteurs constatent aussi la sup eriorit e de leur m ethode dans les p eriodes o u l' etablissement de la correspondance entre les param etres fondamentaux et les cours est moins convaincante. Les auteurs proposent donc une m ethode in edite et simple pour etablir la correspondance entre les param etres comptables fondame...