We investigate American options in a multiple prior setting of continuous

time and determine optimal exercise strategies form the perspective

of an ambiguity averse buyer. The multiple prior setting

relaxes the presumption of a known distribution of the stock price

process and captures the idea of incomplete information of the market

data leading to model uncertainty. Using the theory of (reflected)

backward stochastic differential equations we are able to solve the optimal

stopping problem under multiple priors and identify the particular

worst-case scenario in terms of the worst-case prior. By means of the

analysis of exotic American options we highlight the main difference

to classical single prior models. This is characterized by a resulting

endogenous dynamic structure of the worst-case scenario generated

by model adjustments of the agent due to particular occurring events

that change the agent’s beliefs.