Time-resolved photoluminescence spectra after nonresonant excitation show a distinct 1s resonance, independent of the existence of bound excitons. A microscopic analysis identifies excitonic and electron-hole plasma contributions. For low temperatures and low densities the excitonic emission is extremely sensitive to even minute optically active exciton populations making it possible to extract a phase diagram for incoherent excitonic populations. 1 For a long time, photoluminescence (PL) at the spectral position of the 1s exciton resonance has been considered as evidence for the existence of excitons. The rise of the 1s PL after nonresonant excitation of a semiconductor was interpreted as buildup of an excitonic population [1,2,3,4,5,6], and the PL decay was used to describe exciton recombination [7,8]. However, recently a microscopic theory predicted that PL at the 1s resonance can also originate from correlated plasma emission [9]. Accordingly, PL at the spectral position of the 1s resonance would not prove the existence of excitons, and previous interpretations may be in question. Indeed, in nonresonantly excited time resolved PL measurements the 1s resonance is developed on a sub-ps timescale at 100 K [10], much faster than any expected exciton formation time.Information about exciton formation can be gained by performing THz experiments [11].However, currently the THz results are inconclusive: Kaindl et al. [12] observed the buildup of the induced absorption corresponding to the excitonic 1s to 2p transition showing excitonic populations with formation times on a rather slow timescale of 100's of ps to ns. They claim to observe a nearly completely excitonic system 1 ns after excitation, while Chari et al. [13] only plasma contributions. Also, THz absorption is sensitive to both dark and bright excitons, and cannot answer if and how excitonic populations influence the PL.Here we address the following questions. Is the 1s PL ever dominated by plasma emission?If so, is it always dominated by plasma emission, i.e. what can we learn about excitonic populations from the 1s PL and nonlinear absorption?After ps continuum excitation, time resolved PL and corresponding probe absorption measurements are performed under identical conditions on a ns timescale. The sample (DBR42) consists of 20 MBE-grown 8 nm In 0.06 Ga 0.94 As quantum wells with 130 nm GaAs barriers; both sides are anti-reflection coated. This indium concentration places the 1s exciton resonance at 1.471 eV at 4 K, avoiding absorption in the bulk GaAs substrate that leads to impurity emission at 1.492 eV from unintentional carbon in the substrate. The results, checked on several other samples including a sample grown in a different MBE system, are insensitive to exciton linewidths or to interfacial or alloy disorder. Single-quantum-well data were noisier but exhibited a similar behavior, excluding significant radiative coupling effects.We excite nonresonantly 13.2 meV above the 1s resonance, into the heavy-hole continuum but below the light-h...