The Census Bureau collects data for the American Community Survey (ACS) using four modes:  mail, Internet, Computer-Assisted Telephone Interview (CATI), and Computer-Assisted Personal Interview (CAPI).  In January 2017, the Census Bureau conducted a split-panel experiment to examine the effects of including cellular telephone numbers in the CATI workload.  In this experiment, the test group included a mixture of both landline and cellular numbers, while the control group included only landline numbers. The research then compared various survey productivity and efficiency measures between the two groups, and looked further into how including cellular numbers might affect various aspects of data collection, such as the demographic makeup of the ACS CATI respondents. This report documents the findings from this experiment.