From simple questions sometimes come less than simple answeres.
With that out of the way, the most reliable spot will be the spot that will consistently provide you with the data stream that holds your interest. So if your interest is in looking for wear material then as Vickie suggests select a point in the system where the system generates some selective seperation for you. Remember you are trending - looking for rates of wear particle increasses.
If you are looking for chemical stability - oxidation and or AN etc then the location doesn't matter as long as there is reasonable agitation in the system.
Define your goals(condition monitoring purpose) - most prominant failure contributor, or anticipated failure contributor. Using that as your criteria will help you to narrow down your sampling point.
regards.....
Alan