The simplest tips I can propose, without knowing the structure you are patching, are to apply a stronger positive pressure, to modify the shape of the tip (are you patching depth cell ?), to use a slower speed to lower the electrode.

You can try to move the pipette very slowly as soon as you are in the target area (~1um /s). Also high positive pressure until you reach the target area and low positive pressure for slow movement in the target area is an important factor.

The basic method is explained here: Current protocols in neuroscience UNIT 6.22,  Whole-Cell Recording In Vivo, Michael R. DeWeese

What's the size of your tip? Reducing the diameter of your pipette tip will increase the probability of coming in contact with just the cell, avoiding including the edge of the cell.

1) apply a very small positive pressure in the pipette (if the pressure it too high, you won't be able to see any change in current; see below)

2) in voltage clamp mode, make a test pulse of + 10 mV lasting 50 ms repeated every 200-300 ms

3) bring the pipette down in the tissue with steps of 2 micrometers

4) read the current trace: this is your eyes. Each time the current drops a bit, this might indicate that you touch a membrane (i.e. increase in input resistance). If it happens, withdraw the pipette. The current should go back to control value. Bring the pipette down again. If the current drops again, then you can consider you are touching a cell.

5) release the positive pressure. The current trace should ideally collapse to 0. If not, try to help it by applying a gentle negative pressure while monitoring the current. Get the feeling that the negative pressure bring the current trace down to 0. 

6) when you have a stable gigaseal, move the holding potential to -60 to -70mV and apply a brief sharp small negative pressure to rupture the membrane. Then you are in whole cell.

If you train with this procedure, you should manage to get nice recordings within 1 or 2 weeks.

good luck

http://www.ncbi.nlm.nih.gov/pubmed/25023316

here you can find some tips... 

Jean-Francois Perrier has given a good detail about how to make blind patch.

I add my following comments:

If you will work on old animal (mouse or rat) brain you need high pressure otherwise you will have more chance to touch big fibres and see the resistance change. High pressure makes you to have chance to record neurons on which the electrode is targeting on its middle not on its edge and therefore you will have high chance to get good and high resistance seal (gigaseal).

A wider (i.e. 100ms) Voltage pulse  is better and to watch the pulse change on a good and high frequency monitor (oscilloscope) is also important.

Of course, most important is to get good brain slice with a lot of living cells (Easy to say but difficulty to achieve!) and to work on a nuclear packed by many targeting cells. Otherwise you will record many your-not-targeting cells.

Finally, if it is allowed, it is better to get a visual patch clamp setup because of the huge advance in this area.

Agree with the above people.  Two more addition: For in vivo recording you need the high positive pressure to be persistent during penetrating the tissue, so normal silica gel tubing is too flexible and harder ones like Tygon is beneficial.  Besides, you should avoid any bleeding.

Dear Ping Kwan,

You can find a detailed description of how to make in vivo patch-clamp recording in anesthetized and awake animals in Pernía-Andrade et al., 2012 and Pernía-Andrade & Jonas, 2014 (check out methods and references).