Providing independent advice to the public and professionals for fifteen years
Human infection is always caused by exposure to the bacteria that have been shed by an infected animal, and in 90% of cases it will be their urine (although infection direct from blood is also possible). Direct transfer, where the urine comes into contact with the patient and enters their bloodstream, is very rare except in accidental exposure when handling infected animals, and the usual route is via water that is subsequently drink, or used for recreation such as swimming.
The bacteria have to physcially enter your bloodstream in order to cause an infection, and as they cannot easily penetrate dry undamaged skin, they can only enter at certain locations - injuries where the skin is broken are the obvious places, but mucous membranes lining the airway, mouth, lungs and female sexual organs are also potential routes - so breathing in or swallowing bacteria is a risk, and leptospirosis can sometimes be spread via sexual intercourse.
Dry unbroken skin is a perfect barrier against the bacteria, but cuts and scrapes need only be tiny for the bacteria to find an entry point. There is also a suggestion that the bacteria can pass through very waterlogged skin (such as when skin is immersed in water for a long time), as the cellular structure of the skin changes slightly. This is still only a theory and we have no cases on file.
The bacteria are not generally airborne, so the only risks for breathing in the infection are where water droplets are being created - such as pressure-washing work or in the spray chambers of some air conditioning plant. Being "generally close" to an infected person or animal will not cause an infection!
It's important to realize that the bacteria are incredibly small, and so even a pinhead sized drop of water can carry millions of them. In theory it only takes one to cause illness, but in reality your body's immune system will attack them to a certain extent and so the chances of illness increase as the volume that enters the body (the 'innoculum') increases.
Obviously this depends on two things - if the water is infected with the bacteria, and if you and said bacteria get in close enough contact!
On average in the developed world (Europe, mainland USA, etc.) about 20% of feral rats carry strains of leptospira that could cause illness in humans. This of course varies locally - in your area you may have 100% rat carriers, or 0% - it just depends on the social lives of the rats in question. It's therefore sensible to assume that on average 10% of all freshwater sites are infectious, with more probability for sites which stand good chances of hosting rats nearby (urban ponds, slow-moving rivers and canals, lakes near farm buildings, etc.) and less of a risk for non-rodent-friendly sites such as rapid flowing highland streams or very large estuaries and river deltas. Obviously any site with a high water throughput (such as a river) is less of a risk than stagnant water, as rodent urine will be diluted by the flow. Any water treated with chlorine or UV-sterilisation will be totally safe. This means that swimming pools, and many municipal water fountains and architectural features, are usually of no risk in terms of leptospirosis.
The chances that being exposed to contaminated water would lead to infection depends on what you do in the water. To become infected you must actually allow water to enter your body, though that could be as simple as through an open cut, or by licking a finger. Swimming is the highest risk activity as there is no way to prevent some ingestion and skin contact, though other activities such as fishing, waterskiing, sailing and kayaking can also present risk. Remember that the bacteria cannot survive in saltwater so there is no possible risk from swimming in the sea, or in tidal regions of rivers where the water is briny.
In general in the developed world people are wary of open water sites from general cleanliness viewpoints, and would not drink from a lake without a very good reason. The chances of infection are therefore quite low, but these statistics hide the fact that in many cases the infection is mild, and goes unreported. Despite only a few thousand cases being reported in the developed world each year, there will be many times more cases which are simply written off as a cold or stomach bug - we estimate the total number of cases in the developed world could be up to 100,000 per year.
In developing countries the risks are greater, as rat populations are more widespread and water use is different. The quantity of untreated water used for washing, bathing and drinking is far higher, and the association between hygiene risks and open water is rarely made. Education in developing countries is the only solution to this issue, as the bacteria and the rats are there to stay.
Preventing human infection - an overview
Cleaning and disinfecting procedures