Constraints have important effects on the evolution of life history strategies, but several difficulties have been encountered in determining constraints empirically. Here we investigate methods for measuring a specific type of constraint known as a trade-off. A trade-off between two traits implies a perfect negative correlation between the traits. Trade-offs may involve more than two traits, however, and pairs of traits involved in such a higher-dimensional trade-off may be positively correlated. If some of the traits involved in a trade-off are omitted from the experimental design, the trade-off may not even be detectable. Direct measures of trade-offs are thus complicated, and indirect means of identifying trade-offs may often provide the only feasible measures.