A distinction is drawn between the quantity-oriented approach to memory that has dominated traditional laboratory research, and the accuracy-oriented approach that is emerging in the study of everyday memory. This distinction is shown to underlie some troubling confusions in the interpretation of empirical findings. In particular, the recall-recognition paradox, which involves the claimed superiority of recall over recognition memory in naturalistic settings, is shown to stem from the common confounding between memory property (quantity vs. accuracy) and 2 other variables that have not generally been distinguished--test format (production vs. selection) and report option (free vs. forced reporting). Three laboratory experiments reveal the fundamentally different roles played by report option and test format in accuracy-based and quantity-based memory research. Implications for memory assessment, metamemory, and the everyday-laboratory controversy are discussed.