This paper looks at patterns in business recovery two years after the 2004 Indian Ocean tsunami using longitudinal household survey data for Aceh and North Sumatra gathered 5–14 months and 17–28 months after the event. The base sample contains 2879 households. Applying a weighted probit model that controls for pre-tsunami characteristics the paper finds that the probability of households discontinuing businesses two years after the event increases significantly by 9.6% points in heavy damage areas compared to no/light damage areas. The paper argues that the differences in business recovery may be due to differences in asset recovery patterns. The tsunami wiped out 2/3rd the value of business assets in heavy damage areas and two years later, stocks were still at only half the pre-tsunami values. For these areas business asset growth was significantly correlated with continuity. Higher business discontinuity in heavy damage areas may also be due to higher losses to family and financial support networks experienced by households in these areas with a third of those that discontinued being in temporary accommodation for 6 months or longer. Increased aid and cash-for-work initiatives are unlikely to have been a disincentive to business continuation. The results suggest that given sufficient reconstruction of public infrastructure, asset recovery through livelihood interventions and formal financial market mechanisms is central to business continuity. It also suggests a closer investigation of households that have been displaced for a longer period as the support they require maybe systematically different to support required by other households.