"Zoning laws" as currently understood are a violation of individual rights, and thus would not be allowed in a proper society. However, I believe there is a legitimate place in a proper society for something that may have some similarities to some aspects of zoning laws. I hesitate to call this "something" a zoning law, because this term has so much baggage associated with it and I would be immediately called down by all my fellow objectivists for daring to support the dreaded zoning law in the slightest.
In short, I believe that laws restricting where activities that carry a significant risk of projecting harm into adjacent properties can or cannot be practiced would be proper. So, for example, a law saying that you cannot build an explosive factory within 100 yards of another parcel of property would probably be fine (assuming it can be objectively shown that the explosives factory poses a significant risk of harm). Whether or not such a law would be termed a "zoning law" I will leave to the semanticians to debate. I will explain and justify this proposition below.
The government has the responsibility to protect individuals’ rights, which can be violated by the initiation of force or fraud. However, as some of the commenters have keyed in on, the initiation of force includes the threat of force. Moreover, as anothony argues, the threat of force does not merely include "your money or you life", but also includes creating a significant risk of harm to others. Now, risks of harm surround us constantly, and it cannot be argued that any and all risks are threats of force. However, when a risk of significant harm is sufficiently likely, it becomes a threat of force and is the proper subject of government action. How significant the risk must be before it becomes a threat of force is a difficult question, which is the purview of the philosophy of law to answer.
Some of the commenters have argued that risks of harm should be dealt with by the government only when the harm is actually realized, and that the response should be compensation of the victim by the perpetrator. However, I see no reason why this should be the case, and many good reasons why it should not be the case. Generally, such a position is based on the belief that it is only the realization of the harm that gives rise to the rights violation, but that the risky action itself does not give rise to the rights violation. Thus, from that perspective, no rights violation has occurred until the victim is injured, and thus no government action can be had until then. However, as noted above, the creation of the significant risk of harm is itself a rights violation, and thus the government need not wait until the harm is realized to act to remedy the rights violation.
Let me give you some thought experiments to illustrate how it is proper for the government to act to stop a significant risk of harm before the harm is actually realized.
First, suppose a man in a crowded pedestrian thoroughfare has a revolver with one bullet loaded and the rest of the chambers empty, and the man proceeds to point the gun at the backs of people’s heads and pull the trigger, spinning the chamber after each trigger pull. This creates a significant risk of harm, as each time he pulls the trigger there is a 17% chance that someone will get seriously injured or killed by the man's actions (assuming 6 chambers). However, it is only a risk of harm. Now, are those of you who don't think the government can act until the harm is actually realized really going to tell me that a police officer who sees this event cannot legitimately use force to stop the man? Or that the officer must wait until someone's head is blown off before he can act? Or that the only proper response is forcing the man to compensate the injured victim (or his survivors if he dies)? Obviously the officer would be justified in acting to stop the man before any harm is actually realized. Moreover, the legislature would be justified in banning such actions by law.
Some may argue that the first example is different from a normal risk creation type situation because the man is pointing the gun at people, and therefore his action is an intentional threat and therefore a crime. However, note that the man need not actually intend to harm the people--after all, there is only a 17% chance that the person would be hurt. Moreover, the people don't know they are being threatened (gun is pointed at the back of the head), so there is no question of creating an apprehension of force in the victim. However, these answers will likely not satisfy the critic, so let’s change the example. Suppose instead that the man is not pointing a gun at people, but instead is doing something else that is extremely dangerous (say, building a bomb using volatile ingredients). Just as with the gun, however, there is a 17% chance that something will go terribly wrong and someone nearby will get injured or killed. Does this really change anything from the first answer? Wouldn't the police officer be just as justified in acting before harm occurs in this example as with the gun (assuming the officer knew the risk of harm posed by the action)?
The critics may respond that the above examples are both different from a zoning-type situation, because the actor was on a public pedestrian thoroughfare, not his own property. I don't see how this changes anything at all, and to illustrate lets modify the example again. Suppose the man is not on a sidewalk building his bomb, but instead is in his 500 square foot apartment on the 5th floor of a densely populated 20 floor apartment building. The risk of something going terribly wrong and injuring/killing his neighbors is still 17%. How does the fact that he is in his own apartment change anything? It does not.
Now, if it is proper for the government to act to stop an action that creates a significant risk of harm, then it is proper for the government to create a law barring that action in the first place. This brings us to the non-zoning-something-that-is-somewhat-like-zoning that I refered to earlier. Some activities, like the bomb building in the example above, only create a significant risk of harm when they are done in a location in which the harm can be projected out to the victims. If the man had built his bomb on his 30 acre ranch, then there would be no significant risk of harm being done to any neighbors. Thus, the government could properly act to stop the man from building the bomb in the apartment building, but not on his ranch. If this is so, then what in the world is so wrong with codifying such distinctions, so that they are explicitly and objectively available for the public--i.e., pass a law saying that you can build bombs on your property only if you are X ft. from your nearest neighbor. There is nothing wrong in principle with such a law. Practically speaking, it doesn't make much sense for the legislature to codify every minute risk that could possibly happen, but that is not because it would be wrong to make such a law, just that it might be a waste of time to codify a fact pattern that doesn't occur very often. However, some things may occur frequently enough to justify being codified.