Plug loads are an increasingly important end-use in commercial office buildings. They currently account for 12-50% of total commercial building energy consumption, and as the efficiencies ofregulated major end-uses, such as space conditioning and lighting systems, continue to increase, plugload energy use is expected to rise. This study evaluates patterns in collected plug load data and the effect of a behavior-based intervention to reduce plug load energy consumption.This project leverages a data collection effort originally funded for a study by the California AirResources Board, where 100 plug load monitoring power strips were installed at individualworkstations in the Franklin Building, an office building in Oakland owned by the UC Office of thePresident (UCOP). Each occupant received one power strip and connected up to four devices to beindividually monitored. For this project, only the labeled devices (desktop, laptop, monitor, task light)are included. An analysis of the collected data reveals a clear distinction between work days and non-work days(weekends and holidays). Overall, the monitored occupants have regular work schedules, turn off theirequipment at the end of the work day, and do not often stay late or come in on the weekends. Desktops consume the most power per person, followed by monitors and then task lights. Laptop power trendswere more difficult to discern because users often disconnect them to work in other locations (that werenot monitored). Desktops demonstrate the widest range of power consumption among the devicesmonitored. During unoccupied periods (overnight and on non-work days), desktops draw the mostpower, followed by laptops. All devices draw more power overnight on work days than over weekendsand holidays, indicating that users are more likely to turn equipment off before a longer break from theoffice. Much of the literature on reducing plug load energy consumption in commercial buildings is focusedon technology-based solutions, such as purchasing new equipment or installing sophisticated controlsto turn off equipment when not in use. The literature on changing occupant behavior to reduce energyuse is focused on residential occupants, however multiple studies show that even when occupants donot pay their own bills and have no financial incentive to save energy, other factors can encouragebehavior change. One such motivating method is by using gamification, or turning an everyday activity into a game to encourage behavior change by making it more fun and interesting.With the help of leadership at UCOP, an online sustainability game, Cool Choices, was initiated in theFall of 2014 and 30 employees signed up to play. Cool Choices encourages occupant behavior changesto save water, energy, and reduce waste; players earn points for each action they complete at work or athome and compete with each other on teams. Survey responses from game participants showed thatplayers were motivated to play because the game looked fun, and because the actions suggested wereeasy to perform. An analysis of the energy impact revealed that because occupants were alreadyengaging in relevant energy saving behaviors (e.g. turning equipment off at the end of the day), therewas limited opportunity for further behavior-based reductions. Using trends identified in the baseline analysis, a simplified plug load model was developed to predictpower consumption based on device type, day type (work day or non-work day), and time step, using aMonte Carlo simulation. The model used day type and time step as proxies for occupancy, so when occupancy was not well predicted by the work day/non-work day dichotomy, the model becameincreasingly unreliable. Even after adding an additional variable (month), the model was still not ableto predict power consumption to an acceptable degree of accuracy per industry standards. The model demonstrated a need for a new, more accurate proxy for occupancy, perhaps based on individualoccupants, rather than devices.