In everyday macroeconomic analysis, businessmen and policymakers monitor many variables in order to assess the current situation of a country’s business cycle. However, making this assessment is extremely difficult, especially on the verge of recessions: does a drop in one or more of these series reveal the beginning of a recession? Or is it a signal of a temporal deceleration? To answer these questions we have constructed a monthly probabilistic coincident index to detect how close we are of a recession in the Peruvian economy using a non-linear Markov-switching model. In the construction of this index, we have explored the informational content of tendency surveys and international economic variables. We find that the index detected with promptness and reliability the recent recession period associated with the international financial crisis even in real-time analysis. However, since it has been developed with information comprising eight years due to limited data availability, its future recession detection capability has yet to endure the test of time.