Real option theory has remained a fringe field; practitioners believe it is not practically applicable in complex real world environments. We show that this view is mistaken. We apply real option theory to a highly complex energy problem with unhedgeable risk, time varying volatilities and endogenous exercise dates (non-European options). Investment decisions in the energy industry are often undertaken sequentially and are sensitive to information on markets and geographic conditions. Information may arrive gradually over time and as a consequence of early stage decisions. Contrary to real option analysis (ROA), standard NPV-based frameworks are unsuitable because they do not allow for the fact that new information may change later stage decisions. We apply the approach to exploitation decisions for a Dutch cluster of gas fields, where gas prices and field reservoir size are the two main sources of uncertainty. Gas price returns show volatility clustering , which we model using a GARCH specification. Reservoir size uncertainty is unhedgeable, which necessitates an approach dealing with incomplete markets. Finally investment decisions can be postponed or delayed, which implies an non-European option setting, for which no analytical solutions exist. Correctly modeling the structure of volatility has a major impact: Option values shrink by 50% if the time varying nature of volatility is ignored. We also show that a high correlation between reservoir size at different locations creates large option values. The non-standard features of our approach have a major impact: option values are large so real options based valuations substantially exceed corresponding NPV calculations.