This paper considers the effects of refinery outages (due to planned turn-arounds or un-planned events) on current petroleum product prices and future refinery investment. Empirical evidence on these relationships is mixed and highly dependent on the size and duration of the outage, the geographic area considered, the level of inventories available at the time of the outage, and the tightness of the market as measured by the capacity utilization rate. Using a de-tailed database of plant-level refinery outages for both upstream and downstream refining units, I estimate the effects of outages on product prices controlling for the crude oil price and the ability of operating plants to respond to the outage. I also consider the effect of current market profitability on the likelihood of planned refinery outages and the effects of high utilization rates and planned maintenance on the prospects for unplanned outages. I then use plant-level capac-ity data to analyze the effects of outages, profitability, and utilization rates on future investment decisions of the refinery. Results based on OLS and probit models show that planned outages tend to occur during the spring and fall and during times of relatively low margins as measured by the crack spread. The length of time since the last plant turn-around is positively associated with future unplanned outages. Price regressions show that atmospheric distillation and catalytic cracking outages are positively associated with gasoline prices and the association is stronger the higher is the uti-lization rate at the time of the outage. The relationship between both upstream and downstream investment and outages is mixed though refiners tend to invest less when nearby plants have made investments in the prior year. While causal relationships between outages, prices, and investment are difficult to estimate due to simultaneity and unobserved variables, these descrip-tive results show that outages are an important factor affecting refined product prices and future refinery investment.