It is found that triplet state lifetimes and intersystem crossing yields of individual molecules embedded in a polymer host at room temperature are not constant in time. The range over which the triplet lifetime of a single molecule varies during long observation times shows a strong similarity with the distribution of lifetime values obtained during short observation times of many individual molecules dispersed in space. The similarity is an elegant manifestation of the ergodic principle of statistical physics. 07.79.Fc, Single molecule microscopy and spectroscopy give new and unique insight in the complex behavior of individual emitters on the nanometer scale [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]. Measurements on individual molecules have two distinct advantages. First, phenomena that are usually hidden due to ensemble averaging can be directly observed, such as spectral and rotational jumps [1][2][3][4][5][6][7][8], photon (anti-)bunching [7,[9][10][11][12][13], and discrete photobleaching. Second, the monitoring of single molecule fluorescence forms a powerful way to probe the dynamics of the local environment on a nanometer scale. Therefore, single molecule detection allows the inhomogeneity of the ensemble to be directly related to the real time dynamics of the heterogeneity of the environment. In contrast, alternative techniques to study the composition of an inhomogeneously broadened ensemble like spectral hole-burning or pump-probe spectroscopies still average over a subset of the ensemble.For the purpose of this Letter, fluorescent molecules are considered as a three-level system. Besides the repetitive transitions between the singlet ground state ͑S 0 ͒ and the lowest singlet excited state ͑S 1 ͒ giving rise to fluorescence, the molecule has a small chance to undergo intersystem crossing (ISC) from S 1 to the lowest excited triplet state ͑T 1 ͒. As long as the T 1 state remains occupied, the S 0 2 S 1 transition does not occur and the fluorescence is interrupted temporarily. After decaying to S 0 the molecule starts fluorescing again. As a result, the fluorescence photons are emitted in bunches separated by dark periods when the molecule is in T 1 . This so-called photon bunching can be investigated in two ways. First, autocorrelation of the time intervals between detected photon counts yields the typical duration of the dark periods and thus the T 1 lifetime [7]. The disadvantage of this indirect method is that it yields only a mean value for the T 1 lifetime. Second, integration of the detected fluorescence photons over time intervals shorter than the duration of the dark periods can identify the time length of each excursion to T 1 in a direct way [11][12][13]. We have used the second method to obtain time-resolved T 1 state dynamics.The first measurements on the T 1 state of individual molecules were performed at cryogenic temperatures [1,[7][8][9]11]. It was found that T 1 lifetimes and intersystem crossing rates could vary among different molecules, which was attributed to loc...