I see quite often on indietalk questions or problems regarding recording levels. There seems to be quite a few posts asking or stating what should be the correct recording levels. As Alcove stated, in the days of location tape recorders levels were important because of the high noise when recording to tape but today we have 24bit digital recording and levels are far less important. What you need to be concentrating on is the SNR (signal to noise ratio)! In fact SNR has always been what you need to concentrate on, the reason recording high levels on tape was important was to increase the SNR.
By way of an example, let's say your signal (the dialogue) peaks at -20dB and let's say the noise floor is at -40dB. The SNR is therefore 20dB. Increasing the gain on your mic pre so the dialogue peaks at say -6dB does not improve the SNR because increasing the gain doesn't just make the dialogue 14dB louder, it makes everything which enters the mic 14dB louder, including the noise floor which would now be at -26dB. There is an added consideration though, whacking the gain up on the mic pre is going to increase the amount of noise generated by the mic pre itself. Therefore, depending on the quality of the mic pre, whacking the gain up may make the dialogue sound louder but could actually decrease (make worse) the SNR!
Having said all of this, your stated your signal (dialogue) is peaking at -40dB which is extremely low and could be a problem: Let's say the SNR is 20dB again, meaning your noise floor is at -60dB. At this low level we are likely to be getting close to the point of the self noise generated by a budget recorder and mic combo and therefore could be noticeably reducing our SNR. So while we need to pay far less attention to our recording levels than we once did, we still still have a window of opportunity to achieve, which is well above the level of self noise of the recorder/mic combo but below the level which causes the mic-pre to reduce the SNR.
You always want to be increasing the SNR entering the mic and there are 3 obvious ways to achieve this: 1. Get the mic closer to the signal source (actor's mouth), 2. Decease the ambient noise and 3. Increase the strength of the source signal (make the actor speak louder).
Removing noise in post is generally not a simple procedure and there is always a price to pay because removing the noise also involves removing parts of the signal which is wanted. As noise varies from location to location and often even varies in the same location from moment to moment, there is no one way and no one best tool to remove it. Sometimes one tool (piece of software) removes the noise better and on another occasion that same piece of software may perform noticeably worse.
G