This paper presents an outlier detection algorithm in the state space framework, implemented in JDemetra+ 3.0, the Eurostat recommended seasonal adjustment software. We also introduce the concept of "seasonality break" which refers to a significant change in the seasonal pattern as a whole and thus differs from "seasonal outliers” commonly corrected for. JDemetra+ 3.0 contains an extended state space framework, allowing large-scale seasonal adjustment and outlier detection with basic structural models, which are here compared to the two main seasonal adjustment algorithms Tramo-Seats and X13-Arima-seats.