It has long been debated whether the object marker in Amharic is a reflex of object agreement (Amberber 1996; Yimam 2006) or a doubled object clitic (Mullen 1986; Yabe 2001). I re-open this debate, evaluating previous argument and developing several new arguments for the object marker being an object clitic. If agreement is analyzed as the Minimalist relation Agree, several diagnostics relating to case, the morphophonology of object markers, and what happens when no object marker is possible, all demonstrate that the object markers are doubled clitics. The paper thus serves as a case study in how to distinguish agreement and clitics using multiple diagnostics (cf. Preminger 2009).