I propose that ellipted constituents in English DP (N' ellipsis), CP (sluicing), and IP (VP ellipsis) are empty, non-NP pronominals, licensed and identified under the same condition that constrains nonarbitrary NP pro; [e] must be governed by an X-0 specified for strong agreement. Strong agreement must be broadened to include features that identify non-NP pro, an empty category that crucially differs from NP pro in also requiring reconstruction for its content to be fully recovered. The analysis explains several ungrammatical ellipsis patterns in DP, CP, and IP, and why ellipsis fails in lexical categories and the functional category DEGP in English. The account is supported by ellipsis patterns in German, French, and Malay.