A field test of cheap talk and two types of certainty calibration in contingent valuation of public lands indicated that cheap talk does not reduce WTP estimates. Use of a ten point certainty calibration scale reduces WTP estimates by about half. However, adjusting for uncertainty using a 'Not Sure' option does not reduce WTP estimates but increases the variance in responses. There may be a conceptual difference between these two ways of accounting for respondents' uncertainty, which may suggest why they provide different WTP value estimates and variances.