A document serving as the starting point for evaluating the inventive merits of an invention should relate to the same or a similar technical problem or, at least, to the same or a closely related technical field as the patent in suit (T 495/91, T 570/91, T 989/93, T 1203/97, T 263/99, T 1634/06).
In T 439/92 the board pointed out that one of the criteria for determining the closest prior art was the problem already stated in the patent. In many cases it was reasonable for there to be a link between this problem and the prior art chosen as being closest. See also in this chapter I.D.4.2.2 "Problem formulated in the patent application as starting point".
In T 698/10 the board held that the closest prior art did not have to disclose all the problems solved by the claimed invention. In particular it did not have to disclose the objective technical problem, which was only determined in the second step of the problem and solution approach based on the technical effect(s) provided by those features distinguishing the invention as claimed from the closest prior art. See also T 638/16, T 1148/15, T 1112/19.
In T 325/93 the board stated that the problem addressed by the application was neither derivable nor indeed recognisable from the disclosure of D2 which, according to the department of first instance and the appellant, represented the closest prior art. As early as T 686/91, another board had observed that a document not mentioning a technical problem which was at least related to that derivable from the patent specification did not normally qualify as the closest prior art for inventive step purposes, however many technical features it may have in common with the subject-matter of the patent concerned (see also T 410/93, T 708/96, T 59/96, T 827/00, T 979/00, T 496/02, T 1666/16).
In T 644/97 the board concluded that a technical problem arising from a "closest prior art" disclosure which was irrelevant to the claimed subject-matter (in the sense that it did not mention a problem that was at least related to that derivable from the patent specification) had a form such that its solution could practically never be obvious, because any attempt by the skilled person to establish a chain of considerations leading in an obvious way to the claimed subject-matter was bound to fail (see also T 792/97 and T 599/03). However, according to the board in T 1408/09, it is not a requirement for obviousness that the starting point in the prior art addresses the same technical problem as the claimed invention.
In T 835/00 D1 did not mention any of the problem aspects addressed by the claimed invention. As a consequence, a technical problem was created which was unrelated to the actual disclosure of D1 but whose solution was then found to be obvious in the light of the disclosure of D2. The board referred to T 686/91 and stated that it was a fatal defect that a prior art disclosure from which no relevant technical problem could be formulated without inappropriate hindsight had been chosen as a starting point for the application of the problem and solution approach, because without such hindsight any attempt to establish a logical chain of considerations which might lead to the claimed invention inevitably ran into difficulties at the start, for want of a relevant identifiable goal or object. If the relevant problem was not derivable from the alleged closest prior art, the measures for its solution were a fortiori not derivable. In other words, the invention was not obvious in the light of such art. See also T 548/03, T 1898/07, T 638/16.
In T 25/13, D4 related to tumble dryers and so came under domestic appliances, a field neither the same nor even closely related to that of the invention, namely automotive technology. For that simple reason, it could not be "the closest prior art". The board, however, held that opponents were essentially free to choose the starting point for assessment and that their choice then had implications for the technical knowledge of the relevant skilled person.