However, the devil is in the detail, and that detail manifested on Monday in the form of draft guidelines from BEREC, the association of the European telecoms regulators who will implement the new rules.

The big news is about zero rating: If, when a customer hits her data cap, an operator blocks or slows down all services except for a particular service, it would be breaking the rules. So even if an operator exempts a certain music-streaming service, for example, from counting towards that cap, it won’t be able to get special treatment once the cap is reached.

This should provide to be a strong disincentive for operators who might be thinking about providing data subscriptions with low caps, so they can make it more attractive for application providers to pay them for special treatment.

Get Data Sheet, Fortune’s technology newsletter.

But what of the idea of zero rating itself? Here, the guidelines are quite nuanced.

The regulators said zero rating would break net neutrality rules if they “materially reduce” customers’ choice in practice. They noted that this isn’t automatically the case every time an operator exempts a certain application from the customer’s data cap, but it could be true depending on the market position of the operators and the application providers.

In other words, if a major operator is offering a market-leading service (such as Facebook (FB) or Spotify) for free while charging for its smaller rivals, that could break the rules. “The market positions should be analyzed in line with competition law principles,” the guidelines read.

Other factors include the “effect on freedom of expression and media pluralism,” and if the zero-rating deals have an “effect on the range and diversity of content and applications” or discourage rivals from entering the market.

Does this clear up the zero-rating issue? Yes and no. There’s no universal rule on the subject and regulators will need to examine each deal on a case-by-case basis. As the quality and resources of European regulators also vary by country, there may be some inconsistency down the line—although that’s what these guidelines are supposed to help avoid.

However, this does all very much look like bad news for operators who want to acquire more customers by offering the most popular applications without charging for the data, or hope to use low caps to strong-arm application providers.

As for the “specialized services” that Internet service providers will be able to offer separately from their main Internet packages, the regulators are going to be keeping a close eye on how necessary this separation really is.

Examples of such services are machine-to-machine communications, or “services responding to a public interest” that demand a particularly high quality of transmission. They might include real-time health services, 4G voice, and some Internet TV services. Corporate virtual private network (VPN) services, which offer private connections between a company’s various offices, may also qualify.

The rules passed last year state that operators can only give special treatment to such services when it is truly necessary to do so, when they aren’t a substitute for regular Internet services, and when prioritizing them won’t mean a worse experience for regular Internet users.

For more on net neutrality, watch our video.

According to the guidelines, regulators will need to verify whether the application can really be provided at the special advertised quality, and that it wouldn’t work as well if offered over a regular connection, like everything else, it will need to be provided over a “logically separated” connection.

Again, this will be a case-by-case assessment, and what’s more, it won’t necessarily be a one-off. The guidelines point out that, as regular Internet connections improve over time, an application that requires special treatment today may not require it in the future.

As with zero rating, there does seem to be some leeway here for operators that strong net neutrality advocates won’t like. However, the guidelines hardly give the operators carte blanche to do what they like.

Incidentally, as I have noted before, network-level ad blocking is not allowed under these new rules.

Perhaps in response to European mobile operators toying with the idea of blocking ads on behalf of their customers, the BEREC guidelines specifically use network-level ad blocking as an example of unacceptable behavior (unless the ads are spreading malware or spyware). They also point out that on-device ad blockers, installed by the users themselves, are still kosher.

BEREC’s draft guidelines are open for public comment for the next six weeks.