When a new technology reaches the market, we typically focus on the want or need that it can fulfill. As the technology becomes a commodity and its market matures, reliability often become a key differentiating factor between competing products. We posit that as broadband capacities continue to improve and users migrate to over-the-Internet services, such as on-demand video and voice-over-IP services, we will see this common pattern emerge for broadband services. In this poster, we present the first study of reliability in broadband networks. Using data collected from residential gateways (via FCC/SamKnows), we study the availability and reliability of fixed-line broadband services across the US. Using natural experiments, we look at the impact of increased network downtime on user network demand. We use traditional metrics (e.g. failure rate, MTBF, MTTR) to quantify broadband services, as well as each ISP's configured DNS. Since the impact of a network outage will depend on when it occurred (e.g. time of day), we compare ISP services by the annual average number of bytes lost, based on typical user demand during periods of network downtime.