Recently, the study of chaos is attracting attention, and a wide range of academic fields is actively involved.On the other hand, Aihara proposed the term "chaos enginee~g" to describe the application of chaos theory for engineering purposes, and its possibilities have been demonstrated. The approaches to the engineering applications of chaos are generally classified, into two groups: (1) creative methods and (2) analytical methods. In the former, chaos conditions, consistent with the particular application, are produced in any case, and this behavior is used in a positive manner. In the latter, a phenomenon is examined to see whether it is chaotic or not, and if it is a deterministic chaos, the latent rules involved are extracted and used for a practical purpose. Applications based on the creative methods include chaos computing, chaotic memory, chaos cipher communications, chaos art, applications using chaos fluctuations, chaos circuits, non-linear engineering systems, etc. Applications using the analytical methods include failure diagnosis, deterministic non-hear shortterm forecasting, identifying chaos, modeling, biochaos, etc. This paper first describes approaches of chaotic diagnosis. Next, trajectory parallel measue method for diagnosis is described. Finally, fault diagnosis is discussed as a prospective industrial application using practical examples.In Japan, in 1991, the Bio-information Application System Study Committee (Chaired by K. Aihara) was organized by the Japan Electronic Industry Development Association. The committee surveyed possible engineering applications of chaos, and a report titled "Questionnaire Survey on the Prospects of Chaos Engineering" was published in 1992 [l]. This report used the word "chaos engineering" for the first time in Japan, and since then, R&D on chaos applications has been more widely implemented.Considering the human centered machine system, fault diagnosis is one of the most important functions and also one of the possible applications in chaos engineering field. The approaches of fault diagnosis based on chaos theory are categorized into two groups, one is focused on the changes of dynam~cs in the observed time series by calculation of the suitable embedding dimension and delay time, the maximum Lyapunov exponent and fractal dimension [2], the other is focused on the amount of mixed random noise.For the latter approach, several measures have been proposed to examine whether an observed time series is random noise or chaos. Recently, Kaplan and Glass have proposed the coarse-grained embedding of time series [3], in which the complexity of trajectories embedded in a state space is calculated statistically. With the Kaplan and Glass' algorithm, the state space should be well divided into hypercubes, so it may be necessary to do comprehensive preparatory work, and to do heavy calculation.Under these circumstances, we propose the Trajectory Parallel Measure method (TPM method) [4]. One of the essential points of this method is, similar to the Kaplan and Glass' algorith...