This paper provides a critical review of the literature on non-market valuation methods to estimate the welfare impact of novel products; it is the first study to assess both observed data- and perception-based methods as non-market valuation methods. Observed databased methods include budgets, regression, mathematical programming, and simulation. Perceptions-based methods include the contingent valuation method, choice-based conjoint analysis and experimental methods. Findings imply that the preferred observed data-based method to estimate the ex ante economic impact of a new technology on the welfare of the farm household is a combination of simulation and mathematical programming. The preferred perception-based method for estimating the ex ante impact of a novel product on the welfare of an economic agent is represented by experimental methods. Findings also imply that observed-data based methods and more specifically mathematical programming are more popular for estimating the ex ante farm-level economic impact of a new technology. On the other hand, perception-based methods are more popular for estimating the economic impact of a novel product for consumers.